2013-04-08 23:59:36 +02:00
|
|
|
#
|
|
|
|
|
# spec file for package slurm
|
|
|
|
|
#
|
2019-01-08 20:05:14 +01:00
|
|
|
# Copyright (c) 2019 SUSE LINUX GmbH, Nuernberg, Germany.
|
2013-04-08 23:59:36 +02:00
|
|
|
#
|
|
|
|
|
# All modifications and additions to the file contributed by third parties
|
|
|
|
|
# remain the property of their copyright owners, unless otherwise agreed
|
|
|
|
|
# upon. The license for this file, and modifications and additions to the
|
|
|
|
|
# file, is the same license as for the pristine package itself (unless the
|
|
|
|
|
# license for the pristine package is not an Open Source License, in which
|
|
|
|
|
# case the license is the MIT License). An "Open Source License" is a
|
|
|
|
|
# license that conforms to the Open Source Definition (Version 1.9)
|
|
|
|
|
# published by the Open Source Initiative.
|
|
|
|
|
|
2019-01-31 12:56:59 +01:00
|
|
|
# Please submit bugfixes or comments via https://bugs.opensuse.org/
|
2013-04-08 23:59:36 +02:00
|
|
|
#
|
|
|
|
|
|
2017-05-30 12:41:04 +02:00
|
|
|
|
2017-11-10 15:19:18 +01:00
|
|
|
# Check file META in sources: update so_version to (API_CURRENT - API_AGE)
|
2018-11-20 18:07:44 +01:00
|
|
|
%define so_version 33
|
2019-01-31 12:56:59 +01:00
|
|
|
%define ver 18.08.5
|
2019-01-31 21:33:20 +01:00
|
|
|
%define dl_ver %{ver}-2
|
2017-11-15 13:49:05 +01:00
|
|
|
# so-version is 0 and seems to be stable
|
|
|
|
|
%define pmi_so 0
|
2017-11-01 18:01:38 +01:00
|
|
|
|
2017-02-11 21:31:48 +01:00
|
|
|
# For anything newer than Leap 42.1 and SLE-12-SP1 build compatible to OpenHPC.
|
2017-10-05 07:13:44 +02:00
|
|
|
%if 0%{suse_version} > 1320 || 0%{?sle_version} >= 120200
|
2017-02-11 21:31:48 +01:00
|
|
|
%define OHPC_BUILD 1
|
|
|
|
|
%endif
|
2016-10-16 21:51:20 +02:00
|
|
|
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?suse_version} >= 1220 || 0%{?sle_version} >= 120000
|
2017-02-02 21:23:02 +01:00
|
|
|
%define with_systemd 1
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2019-02-01 20:44:10 +01:00
|
|
|
%if 0%{?suse_version:1} && 0%{?suse_version} <= 1140
|
|
|
|
|
%define comp_at %defattr(-,root,root)
|
|
|
|
|
%endif
|
2017-02-02 21:23:02 +01:00
|
|
|
|
|
|
|
|
%if 0
|
|
|
|
|
%define have_netloc 1
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%ifarch x86_64
|
|
|
|
|
%define have_libnuma 1
|
|
|
|
|
%else
|
|
|
|
|
%ifarch %{ix86}
|
|
|
|
|
%if 0%{?sle_version} >= 120200
|
|
|
|
|
%define have_libnuma 1
|
|
|
|
|
%endif
|
|
|
|
|
%endif
|
2016-10-16 21:51:20 +02:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2017-02-11 21:31:48 +01:00
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%define slurm_u %name
|
|
|
|
|
%define slurm_g %name
|
|
|
|
|
%else
|
|
|
|
|
%define slurm_u daemon
|
|
|
|
|
%define slurm_g root
|
|
|
|
|
%endif
|
|
|
|
|
|
2017-11-10 15:19:18 +01:00
|
|
|
%define libslurm libslurm%{so_version}
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
Name: slurm
|
2017-11-01 18:01:38 +01:00
|
|
|
Version: %{ver}
|
2013-04-08 23:59:36 +02:00
|
|
|
Release: 0
|
|
|
|
|
Summary: Simple Linux Utility for Resource Management
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
License: SUSE-GPL-2.0-with-openssl-exception
|
2013-04-08 23:59:36 +02:00
|
|
|
Group: Productivity/Clustering/Computing
|
2018-03-28 10:20:56 +02:00
|
|
|
Url: https://www.schedmd.com
|
2019-01-31 21:33:20 +01:00
|
|
|
Source: https://download.schedmd.com/slurm/%{name}-%{dl_ver}.tar.bz2
|
2018-03-15 13:23:19 +01:00
|
|
|
Source1: slurm-rpmlintrc
|
2013-04-08 23:59:36 +02:00
|
|
|
Patch0: slurm-2.4.4-rpath.patch
|
|
|
|
|
Patch1: slurm-2.4.4-init.patch
|
2018-03-15 08:03:02 +01:00
|
|
|
Patch2: pam_slurm-Initialize-arrays-and-pass-sizes.patch
|
2018-06-11 12:31:14 +02:00
|
|
|
Patch3: split-xdaemon-in-xdaemon_init-and-xdaemon_finish-for.patch
|
2018-04-12 18:42:36 +02:00
|
|
|
Patch4: slurmctld-uses-xdaemon_-for-systemd.patch
|
|
|
|
|
Patch5: slurmd-uses-xdaemon_-for-systemd.patch
|
|
|
|
|
Patch6: slurmdbd-uses-xdaemon_-for-systemd.patch
|
2018-06-11 12:31:14 +02:00
|
|
|
Patch7: slurmsmwd-uses-xdaemon_-for-systemd.patch
|
|
|
|
|
Patch8: removed-deprecated-xdaemon.patch
|
2018-08-14 15:00:16 +02:00
|
|
|
Patch9: slurmctld-rerun-agent_init-when-backup-controller-takes-over.patch
|
2018-12-12 10:28:26 +01:00
|
|
|
Patch10: pam_slurm_adopt-avoid-running-outside-of-the-sshd-PA.patch
|
|
|
|
|
Patch11: pam_slurm_adopt-send_user_msg-don-t-copy-undefined-d.patch
|
|
|
|
|
Patch12: pam_slurm_adopt-use-uid-to-determine-whether-root-is.patch
|
2018-06-11 12:31:14 +02:00
|
|
|
|
2018-01-30 17:25:18 +01:00
|
|
|
Requires: slurm-config = %{version}
|
|
|
|
|
Requires: slurm-node = %{version}
|
2018-11-20 18:07:44 +01:00
|
|
|
Recommends: slurm-doc = %{version}
|
|
|
|
|
Recommends: slurm-config-man = %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
BuildRequires: fdupes
|
|
|
|
|
BuildRequires: gcc-c++
|
|
|
|
|
BuildRequires: gtk2-devel
|
|
|
|
|
BuildRequires: libbitmask-devel
|
|
|
|
|
BuildRequires: libcpuset-devel
|
2017-10-05 07:13:44 +02:00
|
|
|
BuildRequires: python
|
2017-02-02 21:23:02 +01:00
|
|
|
%if 0%{?have_libnuma}
|
2013-04-08 23:59:36 +02:00
|
|
|
BuildRequires: libnuma-devel
|
|
|
|
|
%endif
|
|
|
|
|
BuildRequires: mysql-devel >= 5.0.0
|
|
|
|
|
BuildRequires: ncurses-devel
|
|
|
|
|
BuildRequires: openssl-devel >= 0.9.6
|
|
|
|
|
BuildRequires: pkgconfig
|
|
|
|
|
BuildRequires: postgresql-devel >= 8.0.0
|
|
|
|
|
BuildRequires: readline-devel
|
2017-02-02 21:23:02 +01:00
|
|
|
%if 0%{?suse_version} > 1310 || 0%{?sle_version}
|
|
|
|
|
BuildRequires: libibmad-devel
|
|
|
|
|
BuildRequires: libibumad-devel
|
|
|
|
|
%endif
|
|
|
|
|
%if 0%{?suse_version} > 1140
|
|
|
|
|
BuildRequires: libhwloc-devel
|
|
|
|
|
%ifarch %{ix86} x86_64
|
|
|
|
|
BuildRequires: freeipmi-devel
|
|
|
|
|
%endif
|
|
|
|
|
%endif
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2014-11-04 21:12:14 +01:00
|
|
|
%{?systemd_requires}
|
2014-11-02 21:27:29 +01:00
|
|
|
BuildRequires: systemd
|
2014-11-04 21:12:14 +01:00
|
|
|
%else
|
2018-09-22 09:50:55 +02:00
|
|
|
Requires(post): %insserv_prereq %fillup_prereq
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
BuildRoot: %{_tmppath}/%{name}-%{version}-build
|
2017-10-05 07:13:44 +02:00
|
|
|
Obsoletes: slurm-sched-wiki < %{version}
|
|
|
|
|
Obsoletes: slurmdb-direct < %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%description
|
2017-10-06 17:44:30 +02:00
|
|
|
SLURM is a fault-tolerant scalable cluster management and job
|
|
|
|
|
scheduling system for Linux clusters containing up to 65,536 nodes.
|
|
|
|
|
Components include machine status, partition management, job
|
|
|
|
|
management, scheduling and accounting modules.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2016-10-16 21:51:20 +02:00
|
|
|
%package doc
|
2017-05-30 12:41:04 +02:00
|
|
|
Summary: Documentation for SLURM
|
|
|
|
|
Group: Documentation/HTML
|
2016-10-16 21:51:20 +02:00
|
|
|
|
|
|
|
|
%description doc
|
2017-10-06 17:44:30 +02:00
|
|
|
Documentation (HTML) for the SLURM cluster managment software.
|
2016-10-16 21:51:20 +02:00
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%package -n perl-slurm
|
|
|
|
|
Summary: Perl API to SLURM
|
|
|
|
|
Group: Development/Languages/Perl
|
|
|
|
|
Requires: slurm = %{version}
|
|
|
|
|
%if 0%{?suse_version} < 1140
|
|
|
|
|
Requires: perl = %{perl_version}
|
|
|
|
|
%else
|
2019-01-31 21:14:27 +01:00
|
|
|
%{libperl_requires}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{perl_requires}
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%description -n perl-slurm
|
2017-10-06 17:44:30 +02:00
|
|
|
This package includes the Perl API to provide an interface to SLURM
|
|
|
|
|
through Perl.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%package -n %{libslurm}
|
2017-10-06 17:44:30 +02:00
|
|
|
Summary: Libraries for SLURM
|
2013-04-08 23:59:36 +02:00
|
|
|
Group: System/Libraries
|
|
|
|
|
|
|
|
|
|
%description -n %{libslurm}
|
|
|
|
|
This package contains the library needed to run programs dynamically linked
|
2017-10-06 17:44:30 +02:00
|
|
|
with SLURM.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
|
2017-11-20 15:01:44 +01:00
|
|
|
%package -n libpmi%{pmi_so}
|
2017-11-10 15:19:18 +01:00
|
|
|
Summary: Libraries for SLURM
|
|
|
|
|
Group: System/Libraries
|
|
|
|
|
|
2017-11-20 15:01:44 +01:00
|
|
|
%description -n libpmi%{pmi_so}
|
2017-11-10 15:19:18 +01:00
|
|
|
This package contains the library needed to run programs dynamically linked
|
|
|
|
|
with SLURM.
|
|
|
|
|
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%package devel
|
|
|
|
|
Summary: Development package for SLURM
|
|
|
|
|
Group: Development/Libraries/C and C++
|
2017-11-15 13:49:05 +01:00
|
|
|
Requires: %{libslurm} = %{version}
|
2017-11-20 15:01:44 +01:00
|
|
|
Requires: libpmi%{pmi_so} = %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
Requires: slurm = %{version}
|
|
|
|
|
|
|
|
|
|
%description devel
|
2017-10-06 17:44:30 +02:00
|
|
|
This package includes the header files for the SLURM API.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
%package auth-none
|
|
|
|
|
Summary: SLURM auth NULL implementation (no authentication)
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
|
|
|
|
Requires: slurm = %{version}
|
|
|
|
|
|
|
|
|
|
%description auth-none
|
|
|
|
|
This package cobtains the SLURM NULL authentication module.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
%package munge
|
|
|
|
|
Summary: SLURM authentication and crypto implementation using Munge
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
2014-07-26 16:36:03 +02:00
|
|
|
Requires: munge
|
2018-01-30 17:25:18 +01:00
|
|
|
Requires: slurm-plugins = %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
BuildRequires: munge-devel
|
|
|
|
|
Obsoletes: slurm-auth-munge < %{version}
|
|
|
|
|
Provides: slurm-auth-munge = %{version}
|
|
|
|
|
|
|
|
|
|
%description munge
|
2018-03-07 21:21:50 +01:00
|
|
|
This package contains the SLURM authentication module for Chris Dunlap's Munge.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%package sview
|
|
|
|
|
Summary: SLURM graphical interface
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
|
|
|
|
|
|
|
|
|
%description sview
|
|
|
|
|
sview is a graphical user interface to get and update state information for
|
|
|
|
|
jobs, partitions, and nodes managed by SLURM.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
%package slurmdbd
|
|
|
|
|
Summary: SLURM database daemon
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
2018-03-28 10:20:56 +02:00
|
|
|
Requires: slurm-config = %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
Requires: slurm-plugins = %{version}
|
2018-04-13 17:08:24 +02:00
|
|
|
Requires: slurm-sql = %{version}
|
2018-08-14 15:00:16 +02:00
|
|
|
Recommends: slurm-munge = %{version}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2014-11-04 21:12:14 +01:00
|
|
|
%{?systemd_requires}
|
|
|
|
|
%else
|
2018-09-22 09:50:55 +02:00
|
|
|
Requires(post): %insserv_prereq %fillup_prereq
|
2014-11-04 21:12:14 +01:00
|
|
|
%endif
|
2017-10-05 07:13:44 +02:00
|
|
|
Obsoletes: slurm-sched-wiki < %{version}
|
|
|
|
|
Obsoletes: slurmdb-direct < %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%description slurmdbd
|
|
|
|
|
The SLURM database daemon provides accounting of jobs in a database.
|
|
|
|
|
|
|
|
|
|
|
2017-10-05 07:13:44 +02:00
|
|
|
%package sql
|
|
|
|
|
Summary: Slurm SQL support
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
|
|
|
|
|
|
|
|
|
%description sql
|
2017-10-06 17:44:30 +02:00
|
|
|
Contains interfaces to MySQL for use by SLURM.
|
2017-10-05 07:13:44 +02:00
|
|
|
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%package plugins
|
|
|
|
|
Summary: SLURM plugins (loadable shared objects)
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
|
|
|
|
|
|
|
|
|
%description plugins
|
|
|
|
|
This package contains the SLURM plugins (loadable shared objects)
|
|
|
|
|
|
|
|
|
|
%package torque
|
2017-10-06 17:44:30 +02:00
|
|
|
Summary: Wrappers for transitition from Torque/PBS to SLURM
|
2013-04-08 23:59:36 +02:00
|
|
|
Group: Productivity/Clustering/Computing
|
2018-09-24 11:37:13 +02:00
|
|
|
Requires: perl-Switch
|
2013-04-08 23:59:36 +02:00
|
|
|
Requires: perl-slurm = %{version}
|
|
|
|
|
Provides: torque-client
|
|
|
|
|
|
|
|
|
|
%description torque
|
2017-10-06 17:44:30 +02:00
|
|
|
Wrapper scripts for aiding migration from Torque/PBS to SLURM.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%package openlava
|
2017-10-06 17:44:30 +02:00
|
|
|
Summary: Wrappers for transitition from OpenLava/LSF to Slurm
|
2017-02-17 13:37:31 +01:00
|
|
|
Group: Productivity/Clustering/Computing
|
2017-05-30 12:41:04 +02:00
|
|
|
Requires: perl-slurm = %{version}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
%description openlava
|
2017-10-06 17:44:30 +02:00
|
|
|
Wrapper scripts for aiding migration from OpenLava/LSF to Slurm
|
2017-02-17 13:37:31 +01:00
|
|
|
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%package seff
|
2017-05-30 12:41:04 +02:00
|
|
|
Summary: Mail tool that includes job statistics in user notification email
|
2017-02-17 13:37:31 +01:00
|
|
|
Group: Productivity/Clustering/Computing
|
2017-05-30 12:41:04 +02:00
|
|
|
Requires: perl-slurm = %{version}
|
2017-02-17 13:37:31 +01:00
|
|
|
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%description seff
|
2017-10-06 17:44:30 +02:00
|
|
|
Mail program used directly by the SLURM daemons. On completion of a job,
|
|
|
|
|
it waits for accounting information to be available and includes that
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
information in the email body.
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%package sjstat
|
|
|
|
|
Summary: Perl tool to print SLURM job state information
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
|
|
|
|
Requires: slurm = %{version}
|
2015-11-01 17:14:37 +01:00
|
|
|
%if 0%{?suse_version} < 1140
|
|
|
|
|
Requires: perl = %{perl_version}
|
|
|
|
|
%else
|
|
|
|
|
%{perl_requires}
|
|
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%description sjstat
|
2017-10-06 17:44:30 +02:00
|
|
|
This package contains a Perl tool to print SLURM job state information.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%package pam_slurm
|
|
|
|
|
Summary: PAM module for restricting access to compute nodes via SLURM
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
2018-12-12 10:28:26 +01:00
|
|
|
Requires: slurm-node = %{version}
|
2013-04-08 23:59:36 +02:00
|
|
|
BuildRequires: pam-devel
|
|
|
|
|
|
|
|
|
|
%description pam_slurm
|
|
|
|
|
This module restricts access to compute nodes in a cluster where the Simple
|
2017-10-06 17:44:30 +02:00
|
|
|
Linux Utility for Resource Managment (SLURM) is in use. Access is granted
|
2013-04-08 23:59:36 +02:00
|
|
|
to root, any user with an SLURM-launched job currently running on the node,
|
2017-10-06 17:44:30 +02:00
|
|
|
or any user who has allocated resources on the node according to the SLURM.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2017-02-02 21:23:02 +01:00
|
|
|
%package lua
|
2017-05-30 12:41:04 +02:00
|
|
|
Summary: Lua API for SLURM
|
2017-02-17 13:37:31 +01:00
|
|
|
Group: Development/Languages/Other
|
2017-02-02 21:23:02 +01:00
|
|
|
Requires: slurm = %{version}
|
|
|
|
|
BuildRequires: lua-devel
|
|
|
|
|
|
|
|
|
|
%description lua
|
2017-10-06 17:44:30 +02:00
|
|
|
This package includes the Lua API to provide an interface to SLURM
|
|
|
|
|
through Lua.
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2018-01-30 17:25:18 +01:00
|
|
|
%package node
|
|
|
|
|
Summary: Minimal slurm node
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
|
|
|
|
Requires: slurm-config = %{version}
|
|
|
|
|
Requires: slurm-plugins = %{version}
|
|
|
|
|
Recommends: %{name}-munge
|
2018-09-22 09:50:55 +02:00
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%{?systemd_requires}
|
|
|
|
|
%else
|
|
|
|
|
Requires(post): %insserv_prereq %fillup_prereq
|
|
|
|
|
%endif
|
2018-01-30 17:25:18 +01:00
|
|
|
|
|
|
|
|
%description node
|
|
|
|
|
This package contains just the minmal code to run a compute node.
|
|
|
|
|
|
|
|
|
|
%package config
|
|
|
|
|
Summary: Config files and directories for slurm services
|
|
|
|
|
Group: Productivity/Clustering/Computing
|
2018-09-22 09:50:55 +02:00
|
|
|
%if 0%{?suse_version} <= 1140
|
|
|
|
|
Requires(pre): pwdutils
|
|
|
|
|
%else
|
|
|
|
|
Requires(pre): shadow
|
|
|
|
|
%endif
|
|
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%{?systemd_requires}
|
|
|
|
|
%endif
|
2018-01-30 17:25:18 +01:00
|
|
|
|
|
|
|
|
%description config
|
|
|
|
|
This package contains the slurm config files necessary direcories
|
|
|
|
|
for the slurm daemons.
|
|
|
|
|
|
2018-11-20 18:07:44 +01:00
|
|
|
%package config-man
|
|
|
|
|
Summary: Config files and directories for slurm services
|
|
|
|
|
Group: Documentation/Man
|
|
|
|
|
|
|
|
|
|
%description config-man
|
|
|
|
|
Man pages for the SLURM cluster managment software config files.
|
2018-01-30 17:25:18 +01:00
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%prep
|
2019-01-31 21:33:20 +01:00
|
|
|
%setup -q -n %{name}-%{dl_ver}
|
2013-04-08 23:59:36 +02:00
|
|
|
%patch0 -p1
|
|
|
|
|
%patch1 -p1
|
2016-10-16 21:51:20 +02:00
|
|
|
%patch2 -p1
|
2018-04-12 18:42:36 +02:00
|
|
|
%patch3 -p1
|
|
|
|
|
%patch4 -p1
|
|
|
|
|
%patch5 -p1
|
|
|
|
|
%patch6 -p1
|
|
|
|
|
%patch7 -p1
|
2018-06-11 12:31:14 +02:00
|
|
|
%patch8 -p1
|
2019-01-31 12:56:59 +01:00
|
|
|
# Drop this fix as it is considered to be resolved by c1a537dbbe6
|
|
|
|
|
##%patch9 -p1
|
2018-12-12 10:28:26 +01:00
|
|
|
%patch10 -p1
|
|
|
|
|
%patch11 -p1
|
|
|
|
|
%patch12 -p1
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%build
|
|
|
|
|
%configure --enable-shared \
|
|
|
|
|
--disable-static \
|
|
|
|
|
--without-rpath \
|
2017-02-02 21:23:02 +01:00
|
|
|
%{!?have_netloc:--without-netloc} \
|
2013-04-08 23:59:36 +02:00
|
|
|
--sysconfdir=%{_sysconfdir}/%{name}
|
|
|
|
|
make %{?_smp_mflags}
|
|
|
|
|
|
|
|
|
|
%install
|
2017-02-02 21:23:02 +01:00
|
|
|
%make_install
|
2017-05-27 15:49:16 +02:00
|
|
|
make install-contrib DESTDIR=%{buildroot} PERL_MM_PARAMS="INSTALLDIRS=vendor"
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
|
|
|
|
%if 0%{?with_systemd}
|
2014-11-02 21:27:29 +01:00
|
|
|
mkdir -p %{buildroot}%{_unitdir}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
install -p -m644 etc/slurmd.service etc/slurmdbd.service etc/slurmctld.service %{buildroot}%{_unitdir}
|
|
|
|
|
ln -s /usr/sbin/service %{buildroot}%{_sbindir}/rcslurmd
|
2014-11-04 21:12:14 +01:00
|
|
|
ln -s /usr/sbin/service %{buildroot}%{_sbindir}/rcslurmdbd
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
ln -s /usr/sbin/service %{buildroot}%{_sbindir}/rcslurmctld
|
2017-10-05 07:13:44 +02:00
|
|
|
install -d -m 0755 %{buildroot}/%{_tmpfilesdir}/
|
|
|
|
|
cat <<-EOF > %{buildroot}/%{_tmpfilesdir}/%{name}.conf
|
|
|
|
|
# Create a directory with permissions 0700 owned by user slurm, group slurm
|
2018-03-07 21:21:50 +01:00
|
|
|
d /var/run/slurm 0700 slurm slurm
|
2017-10-05 07:13:44 +02:00
|
|
|
EOF
|
|
|
|
|
chmod 0644 %{buildroot}/%{_tmpfilesdir}/%{name}.conf
|
2014-11-02 21:27:29 +01:00
|
|
|
%else
|
2017-05-27 15:49:16 +02:00
|
|
|
install -D -m755 etc/init.d.slurm %{buildroot}%{_initrddir}/slurm
|
|
|
|
|
install -D -m755 etc/init.d.slurmdbd %{buildroot}%{_initrddir}/slurmdbd
|
2013-04-08 23:59:36 +02:00
|
|
|
ln -sf %{_initrddir}/slurm %{buildroot}%{_sbindir}/rcslurm
|
|
|
|
|
ln -sf %{_initrddir}/slurmdbd %{buildroot}%{_sbindir}/rcslurmdbd
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2017-12-07 12:18:20 +01:00
|
|
|
mkdir -p %{buildroot}%{_var}/spool/slurm
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2017-10-05 07:13:44 +02:00
|
|
|
rm -f contribs/cray/opt_modulefiles_slurm
|
|
|
|
|
rm -f %{buildroot}%{_sysconfdir}/plugstack.conf.template
|
|
|
|
|
rm -f %{buildroot}%{_sysconfdir}/slurm.conf.template
|
|
|
|
|
rm -f %{buildroot}%{_sbindir}/capmc_suspend
|
|
|
|
|
rm -f %{buildroot}%{_sbindir}/capmc_resume
|
|
|
|
|
rm -f %{buildroot}%{_sbindir}/slurmconfgen.py
|
|
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
install -D -m644 etc/cgroup.conf.example %{buildroot}/%{_sysconfdir}/%{name}/cgroup.conf
|
2017-10-05 07:13:44 +02:00
|
|
|
install -D -m644 etc/layouts.d.power.conf.example %{buildroot}/%{_sysconfdir}/%{name}/layouts.d/power.conf.example
|
|
|
|
|
install -D -m644 etc/layouts.d.power_cpufreq.conf.example %{buildroot}/%{_sysconfdir}/%{name}/layouts.d/power_cpufreq.conf.example
|
|
|
|
|
install -D -m644 etc/layouts.d.unit.conf.example %{buildroot}/%{_sysconfdir}/%{name}/layouts.d/unit.conf.example
|
|
|
|
|
install -D -m644 etc/slurm.conf.example %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf%{?OHPC_BUILD:.example}
|
|
|
|
|
install -D -m644 etc/slurmdbd.conf.example %{buildroot}/%{_sysconfdir}/%{name}/slurmdbd.conf
|
|
|
|
|
install -D -m644 etc/slurmdbd.conf.example %{buildroot}%{_sysconfdir}/%{name}/slurmdbd.conf.example
|
2017-05-27 15:49:16 +02:00
|
|
|
install -D -m755 contribs/sjstat %{buildroot}%{_bindir}/sjstat
|
2017-10-05 07:13:44 +02:00
|
|
|
install -D -m755 contribs/sgather/sgather %{buildroot}%{_bindir}/sgather
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
2018-04-12 19:22:25 +02:00
|
|
|
cp contribs/pam_slurm_adopt/README ../README.pam_slurm_adopt
|
|
|
|
|
cp contribs/pam/README ../README.pam_slurm
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?OHPC_BUILD}
|
2018-03-28 10:20:56 +02:00
|
|
|
# change slurm.conf for our needs
|
2017-02-17 13:37:31 +01:00
|
|
|
head -n -2 %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf.example | grep -v ReturnToService > %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf
|
|
|
|
|
sed -i 's#\(StateSaveLocation=\).*#\1%_localstatedir/lib/slurm#' %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf
|
2017-10-05 07:13:44 +02:00
|
|
|
sed -i 's#^\(SlurmdPidFile=\).*$#\1%{_localstatedir}/run/slurm/slurmd.pid#' %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf
|
|
|
|
|
sed -i 's#^\(SlurmctldPidFile=\).*$#\1%{_localstatedir}/run/slurm/slurmctld.pid#' %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf
|
2017-12-07 12:18:20 +01:00
|
|
|
sed -i 's#^\(SlurmdSpoolDir=\)/.*#\1%{_localstatedir}/spool/slurm#' %{buildroot}/%{_sysconfdir}/%{name}/slurm.conf
|
|
|
|
|
cat >>%{buildroot}/%{_sysconfdir}/%{name}/slurm.conf <<EOF
|
|
|
|
|
# SUSE default configuration
|
|
|
|
|
PropagateResourceLimitsExcept=MEMLOCK
|
|
|
|
|
NodeName=linux State=UNKNOWN
|
|
|
|
|
PartitionName=normal Nodes=linux Default=YES MaxTime=24:00:00 State=UP
|
|
|
|
|
EOF
|
2018-08-14 15:00:16 +02:00
|
|
|
# change slurmdbd.conf for our needs
|
|
|
|
|
sed -i 's@LogFile=/var/log/slurm/slurmdbd.log@LogFile=/var/log/slurmdbd.log@'\
|
|
|
|
|
%{buildroot}/%{_sysconfdir}/%{name}/slurmdbd.conf
|
|
|
|
|
sed -i -e "s@PidFile=.*@PidFile=%{_localstatedir}/run/slurm/slurmdbd.pid@" \
|
|
|
|
|
%{buildroot}/%{_sysconfdir}/%{name}/slurmdbd.conf
|
|
|
|
|
# manage local state dir and a remote states save location
|
2017-02-17 13:37:31 +01:00
|
|
|
mkdir -p %{buildroot}/%_localstatedir/lib/slurm
|
2018-08-14 15:00:16 +02:00
|
|
|
sed -i -e "s@PIDFile=.*@PIDFile=%{_localstatedir}/run/slurm/slurmctld.pid@" \
|
|
|
|
|
-e "s@After=.*@After=network.target munge.service remote-fs.target@" \
|
|
|
|
|
%{buildroot}/%{_unitdir}/slurmctld.service
|
|
|
|
|
sed -i -e "s@PIDFile=.*@PIDFile=%{_localstatedir}/run/slurm/slurmd.pid@" \
|
|
|
|
|
%{buildroot}/%{_unitdir}/slurmd.service
|
|
|
|
|
sed -i -e "s@PIDFile=.*@PIDFile=%{_localstatedir}/run/slurm/slurmdbd.pid@" \
|
|
|
|
|
%{buildroot}/%{_unitdir}/slurmdbd.service
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
# Delete unpackaged files:
|
2017-10-05 07:13:44 +02:00
|
|
|
test -s %{buildroot}/%{_perldir}/auto/Slurm/Slurm.bs ||
|
|
|
|
|
rm -f %{buildroot}/%{_perldir}/auto/Slurm/Slurm.bs
|
|
|
|
|
|
|
|
|
|
test -s %{buildroot}/%{_perldir}/auto/Slurmdb/Slurmdb.bs ||
|
|
|
|
|
rm -f %{buildroot}/%{_perldir}/auto/Slurmdb/Slurmdb.bs
|
|
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
rm -rf %{buildroot}/%{_libdir}/slurm/*.{a,la} \
|
|
|
|
|
%{buildroot}/%{_libdir}/*.la \
|
|
|
|
|
%{buildroot}/%_lib/security/*.la \
|
|
|
|
|
%{buildroot}/%{_mandir}/man5/bluegene*
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
rm -f %{buildroot}/%{_mandir}/man1/srun_cr* \
|
|
|
|
|
%{buildroot}/%{_bindir}/srun_cr \
|
|
|
|
|
%{buildroot}/%{_libexecdir}/slurm/cr_*
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
rm doc/html/shtml2html.py doc/html/Makefile*
|
|
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
rm -f %{buildroot}/%{perl_archlib}/perllocal.pod
|
|
|
|
|
rm -f %{buildroot}/%{perl_vendorarch}/auto/Slurm/.packlist
|
|
|
|
|
rm -f %{buildroot}/%{perl_vendorarch}/auto/Slurmdb/.packlist
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
|
|
|
|
# Build man pages that are generated directly by the tools
|
2017-02-17 13:37:31 +01:00
|
|
|
rm -f %{buildroot}/%{_mandir}/man1/sjobexitmod.1
|
2017-05-27 15:49:16 +02:00
|
|
|
%{buildroot}%{_bindir}/sjobexitmod --roff > %{buildroot}/%{_mandir}/man1/sjobexitmod.1
|
2017-02-17 13:37:31 +01:00
|
|
|
rm -f %{buildroot}/%{_mandir}/man1/sjstat.1
|
2017-05-27 15:49:16 +02:00
|
|
|
%{buildroot}%{_bindir}/sjstat --roff > %{buildroot}/%{_mandir}/man1/sjstat.1
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
2015-11-01 17:14:37 +01:00
|
|
|
# rpmlint reports wrong end of line for those files
|
2017-02-17 13:37:31 +01:00
|
|
|
sed -i 's/\r$//' %{buildroot}/%{_bindir}/qrerun
|
|
|
|
|
sed -i 's/\r$//' %{buildroot}/%{_bindir}/qalter
|
2015-11-01 17:14:37 +01:00
|
|
|
|
2017-05-30 12:41:04 +02:00
|
|
|
# avoid conflicts with other packages, make wrapper unique
|
|
|
|
|
mv %{buildroot}/%{_bindir}/mpiexec %{buildroot}/%{_bindir}/mpiexec.slurm
|
|
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
mkdir -p %{buildroot}/etc/ld.so.conf.d
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
echo '%{_libdir}
|
2017-02-17 13:37:31 +01:00
|
|
|
%{_libdir}/slurm' > %{buildroot}/etc/ld.so.conf.d/slurm.conf
|
|
|
|
|
chmod 644 %{buildroot}/etc/ld.so.conf.d/slurm.conf
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
|
|
|
|
# Make pkg-config file
|
2017-02-17 13:37:31 +01:00
|
|
|
mkdir -p %{buildroot}/%{_libdir}/pkgconfig
|
|
|
|
|
cat > %{buildroot}/%{_libdir}/pkgconfig/slurm.pc <<EOF
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
includedir=%{_prefix}/include
|
|
|
|
|
libdir=%{_libdir}
|
|
|
|
|
|
|
|
|
|
Cflags: -I\${includedir}
|
|
|
|
|
Libs: -L\${libdir} -lslurm
|
|
|
|
|
Description: Slurm API
|
2017-02-02 21:23:02 +01:00
|
|
|
Name: %{name}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
Version: %{version}
|
|
|
|
|
EOF
|
|
|
|
|
|
2018-12-12 10:28:26 +01:00
|
|
|
# Enable rotation of log files
|
|
|
|
|
|
|
|
|
|
mkdir -p %{buildroot}/%{_sysconfdir}/logrotate.d/
|
|
|
|
|
for service in slurmd slurmctld slurmdbd ; do
|
|
|
|
|
cat <<EOF > %{buildroot}/%{_sysconfdir}/logrotate.d/${service}.conf
|
|
|
|
|
/var/log/${service}.log {
|
|
|
|
|
compress
|
|
|
|
|
dateext
|
|
|
|
|
missingok
|
|
|
|
|
nocreate
|
|
|
|
|
notifempty
|
|
|
|
|
maxage 365
|
|
|
|
|
rotate 99
|
|
|
|
|
copytruncate
|
|
|
|
|
postrotate
|
|
|
|
|
pgrep ${service} && killall -SIGUSR2 ${service} || exit 0
|
|
|
|
|
endscript
|
|
|
|
|
}
|
|
|
|
|
EOF
|
|
|
|
|
done
|
|
|
|
|
|
2017-02-17 13:37:31 +01:00
|
|
|
%fdupes -s %{buildroot}
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2014-11-02 21:27:29 +01:00
|
|
|
%pre
|
2017-02-02 21:23:02 +01:00
|
|
|
%if 0%{?with_systemd}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%service_add_pre slurmctld.service
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%post
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%service_add_post slurmctld.service
|
2014-11-02 21:27:29 +01:00
|
|
|
%else
|
2013-04-08 23:59:36 +02:00
|
|
|
%fillup_and_insserv slurm
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%preun
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%service_del_preun slurmctld.service
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%postun
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2018-04-11 13:50:15 +02:00
|
|
|
%service_del_postun -n slurmctld.service
|
2014-11-02 21:27:29 +01:00
|
|
|
%else
|
2013-04-08 23:59:36 +02:00
|
|
|
%insserv_cleanup
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
|
|
|
|
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2014-11-02 21:27:29 +01:00
|
|
|
%pre slurmdbd
|
|
|
|
|
%service_add_pre slurmdbd.service
|
|
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%post slurmdbd
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2014-11-02 21:27:29 +01:00
|
|
|
%service_add_post slurmdbd.service
|
|
|
|
|
%else
|
2013-04-08 23:59:36 +02:00
|
|
|
%fillup_and_insserv slurmdbd
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%preun slurmdbd
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2014-11-02 21:27:29 +01:00
|
|
|
%service_del_preun slurmdbd.service
|
|
|
|
|
%else
|
2013-04-08 23:59:36 +02:00
|
|
|
%stop_on_removal slurmdbd
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%postun slurmdbd
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%if 0%{?with_systemd}
|
2018-04-11 13:50:15 +02:00
|
|
|
%service_del_postun -n slurmdbd.service
|
2014-11-02 21:27:29 +01:00
|
|
|
%else
|
2013-04-08 23:59:36 +02:00
|
|
|
%restart_on_update slurmdbd
|
|
|
|
|
%insserv_cleanup
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2018-01-30 17:25:18 +01:00
|
|
|
%pre node
|
|
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%service_add_pre slurmd.service
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%post node
|
|
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%service_add_post slurmd.service
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%preun node
|
|
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%service_del_preun slurmd.service
|
|
|
|
|
%else
|
|
|
|
|
%stop_on_removal slurmd
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%postun node
|
|
|
|
|
%if 0%{?with_systemd}
|
2018-04-11 13:50:15 +02:00
|
|
|
%service_del_postun -n slurmd.service
|
2018-01-30 17:25:18 +01:00
|
|
|
%else
|
|
|
|
|
%restart_on_update slurmd
|
|
|
|
|
%insserv_cleanup
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%pre config
|
|
|
|
|
%define slurmdir %{_sysconfdir}/slurm
|
|
|
|
|
%define slurmdescr "SLURM workload manager"
|
|
|
|
|
getent group %slurm_g >/dev/null || groupadd -r %slurm_g
|
|
|
|
|
getent passwd %slurm_u >/dev/null || useradd -r -g %slurm_g -d %slurmdir -s /bin/false -c %{slurmdescr} %slurm_u
|
|
|
|
|
exit 0
|
|
|
|
|
|
|
|
|
|
%post config
|
|
|
|
|
%if 0%{?with_systemd}
|
2018-03-15 13:23:19 +01:00
|
|
|
%if 0%{?tmpfiles_create:1}
|
2018-01-30 17:25:18 +01:00
|
|
|
%tmpfiles_create slurm.conf
|
|
|
|
|
%else
|
|
|
|
|
systemd-tmpfiles --create slurm.conf
|
|
|
|
|
%endif
|
|
|
|
|
%endif
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%post -n %{libslurm} -p /sbin/ldconfig
|
|
|
|
|
%postun -n %{libslurm} -p /sbin/ldconfig
|
|
|
|
|
|
2017-11-20 15:01:44 +01:00
|
|
|
%post -n libpmi%{pmi_so} -p /sbin/ldconfig
|
|
|
|
|
%postun -n libpmi%{pmi_so} -p /sbin/ldconfig
|
2017-11-10 15:19:18 +01:00
|
|
|
|
2018-07-11 14:08:06 +02:00
|
|
|
%{!?nil:
|
2018-04-11 13:50:15 +02:00
|
|
|
# On update the %%postun code of the old package restarts the
|
|
|
|
|
# service. This breaks in case the ABI between slurm and its
|
|
|
|
|
# plugins has changed as updates are not atomic. Since we cannot
|
|
|
|
|
# fix the old scripts we need these macros as a workaround.
|
|
|
|
|
# They should be removed at some point.
|
2018-04-20 11:24:13 +02:00
|
|
|
# Do pretrans in lua: https://fedoraproject.org/wiki/Packaging:Scriptlets
|
2018-07-11 14:08:06 +02:00
|
|
|
}
|
2018-04-20 11:24:13 +02:00
|
|
|
%define _test_rest() %{?with_systemd: os.remove("/run/%{1}.rst")
|
|
|
|
|
if os.execute() and os.getenv("YAST_IS_RUNNING") ~= "instsys" then
|
2018-12-12 10:28:26 +01:00
|
|
|
local handle = io.popen("systemctl is-active %{1} 2>&1")
|
2018-12-06 10:50:36 +01:00
|
|
|
local str = handle:read("*a"); handle:close()
|
|
|
|
|
str = string.gsub(str, '^%%s+', '')
|
|
|
|
|
str = string.gsub(str, '%%s+$', '')
|
|
|
|
|
str = string.gsub(str, '[\\n\\r]+', ' ')
|
2018-12-12 10:28:26 +01:00
|
|
|
if str == "active" then
|
2018-12-06 10:50:36 +01:00
|
|
|
local file = io.open("/run/%{1}.rst","w"); file:close()
|
2018-04-20 11:24:13 +02:00
|
|
|
end
|
|
|
|
|
end
|
|
|
|
|
}
|
2018-04-11 13:50:15 +02:00
|
|
|
%define _rest() %{?with_systemd:[ -e /run/%{1}.rst ] && { systemctl status %{1} &>/dev/null || systemctl restart %{1}; }; rm -f /run/%{1}.rst;}
|
2018-07-11 14:08:06 +02:00
|
|
|
%{!?nil:
|
2018-04-11 13:50:15 +02:00
|
|
|
# Until a posttrans macro has been added to macros.systemd, we need this
|
2018-07-11 14:08:06 +02:00
|
|
|
}
|
2018-04-11 13:50:15 +02:00
|
|
|
%define _res_update() %{?with_systemd:%{expand:%%_restart_on_update %{?*}};}
|
|
|
|
|
|
2018-04-20 11:24:13 +02:00
|
|
|
%pretrans -p <lua>
|
2018-04-11 13:50:15 +02:00
|
|
|
%_test_rest slurmctld
|
|
|
|
|
|
2018-04-20 11:24:13 +02:00
|
|
|
%pretrans node -p <lua>
|
2018-04-11 13:50:15 +02:00
|
|
|
%_test_rest slurmd
|
|
|
|
|
|
2018-04-20 11:24:13 +02:00
|
|
|
%pretrans slurmdbd -p <lua>
|
2018-04-11 13:50:15 +02:00
|
|
|
%_test_rest slurmdbd
|
|
|
|
|
|
|
|
|
|
%posttrans
|
|
|
|
|
%_res_update slurmctld
|
|
|
|
|
%_rest slurmctld
|
|
|
|
|
|
|
|
|
|
%posttrans node
|
|
|
|
|
%_res_update slurmd
|
|
|
|
|
%_rest slurmd
|
|
|
|
|
|
|
|
|
|
%posttrans slurmdbd
|
|
|
|
|
%_res_update slurmdbd.service
|
|
|
|
|
%_rest slurmdbd
|
|
|
|
|
|
2018-08-14 15:00:16 +02:00
|
|
|
%if 0%{?sle_version} > 120200 || 0%{?suse_version} > 1320
|
|
|
|
|
%define my_license %license
|
|
|
|
|
%else
|
|
|
|
|
%define my_license %doc
|
|
|
|
|
%endif
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%files
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2018-04-12 19:22:25 +02:00
|
|
|
%doc AUTHORS NEWS RELEASE_NOTES DISCLAIMER
|
2018-08-14 15:00:16 +02:00
|
|
|
%my_license COPYING
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/sacct
|
|
|
|
|
%{_bindir}/sacctmgr
|
|
|
|
|
%{_bindir}/salloc
|
|
|
|
|
%{_bindir}/sattach
|
|
|
|
|
%{_bindir}/sbatch
|
|
|
|
|
%{_bindir}/sbcast
|
|
|
|
|
%{_bindir}/scancel
|
|
|
|
|
%{_bindir}/scontrol
|
|
|
|
|
%{_bindir}/sdiag
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_bindir}/sgather
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/sinfo
|
|
|
|
|
%{_bindir}/sjobexitmod
|
|
|
|
|
%{_bindir}/sprio
|
|
|
|
|
%{_bindir}/squeue
|
|
|
|
|
%{_bindir}/sreport
|
|
|
|
|
%{_bindir}/srun
|
|
|
|
|
%{_bindir}/smap
|
|
|
|
|
%{_bindir}/sshare
|
|
|
|
|
%{_bindir}/sstat
|
|
|
|
|
%{_bindir}/strigger
|
2017-02-11 21:31:48 +01:00
|
|
|
%{?have_netloc:%{_bindir}/netloc_to_topology}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_sbindir}/slurmctld
|
2018-06-11 12:31:14 +02:00
|
|
|
%{_sbindir}/slurmsmwd
|
2018-11-20 18:07:44 +01:00
|
|
|
%dir %{_libdir}/slurm/src
|
|
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%{_unitdir}/slurmctld.service
|
|
|
|
|
%{_sbindir}/rcslurmctld
|
2019-01-31 21:14:27 +01:00
|
|
|
%endif
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_mandir}/man1/sacct.1*
|
|
|
|
|
%{_mandir}/man1/sacctmgr.1*
|
|
|
|
|
%{_mandir}/man1/salloc.1*
|
|
|
|
|
%{_mandir}/man1/sattach.1*
|
|
|
|
|
%{_mandir}/man1/sbatch.1*
|
|
|
|
|
%{_mandir}/man1/sbcast.1*
|
|
|
|
|
%{_mandir}/man1/scancel.1*
|
|
|
|
|
%{_mandir}/man1/scontrol.1*
|
|
|
|
|
%{_mandir}/man1/sdiag.1.*
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_mandir}/man1/sgather.1.*
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_mandir}/man1/sinfo.1*
|
|
|
|
|
%{_mandir}/man1/slurm.1*
|
|
|
|
|
%{_mandir}/man1/smap.1*
|
|
|
|
|
%{_mandir}/man1/sprio.1*
|
|
|
|
|
%{_mandir}/man1/squeue.1*
|
|
|
|
|
%{_mandir}/man1/sreport.1*
|
|
|
|
|
%{_mandir}/man1/srun.1*
|
|
|
|
|
%{_mandir}/man1/sshare.1*
|
|
|
|
|
%{_mandir}/man1/sstat.1*
|
|
|
|
|
%{_mandir}/man1/strigger.1*
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%{_mandir}/man1/sjobexitmod.1.*
|
|
|
|
|
%{_mandir}/man1/sjstat.1.*
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_mandir}/man8/slurmctld.*
|
|
|
|
|
%{_mandir}/man8/spank*
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
|
|
|
|
|
%files openlava
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%{_bindir}/bjobs
|
|
|
|
|
%{_bindir}/bkill
|
|
|
|
|
%{_bindir}/bsub
|
|
|
|
|
%{_bindir}/lsid
|
|
|
|
|
|
|
|
|
|
%files seff
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%{_bindir}/seff
|
|
|
|
|
%{_bindir}/smail
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2016-10-16 21:51:20 +02:00
|
|
|
%files doc
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2019-01-31 22:19:18 +01:00
|
|
|
%dir %{_datadir}/doc/%{name}-%{dl_ver}
|
|
|
|
|
%{_datadir}/doc/%{name}-%{dl_ver}/*
|
2016-10-16 21:51:20 +02:00
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%files -n %{libslurm}
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2017-11-20 15:01:44 +01:00
|
|
|
%{_libdir}/libslurm*.so.%{so_version}*
|
2017-11-10 15:19:18 +01:00
|
|
|
|
2017-11-20 15:01:44 +01:00
|
|
|
%files -n libpmi%{pmi_so}
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2017-11-20 15:01:44 +01:00
|
|
|
%{_libdir}/libpmi*.so.%{pmi_so}*
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%files devel
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_prefix}/include/slurm
|
2017-02-02 21:23:02 +01:00
|
|
|
%{_libdir}/libpmi.so
|
|
|
|
|
%{_libdir}/libpmi2.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/libslurm.so
|
|
|
|
|
%{_libdir}/libslurmdb.so
|
|
|
|
|
%{_libdir}/slurm/src/*
|
|
|
|
|
%{_mandir}/man3/slurm_*
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%{_libdir}/pkgconfig/slurm.pc
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%files sview
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/sview
|
|
|
|
|
%{_mandir}/man1/sview.1*
|
|
|
|
|
|
|
|
|
|
%files auth-none
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/auth_none.so
|
|
|
|
|
|
|
|
|
|
%files munge
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/auth_munge.so
|
|
|
|
|
%{_libdir}/slurm/crypto_munge.so
|
|
|
|
|
|
|
|
|
|
%files -n perl-slurm
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{perl_vendorarch}/Slurm.pm
|
|
|
|
|
%{perl_vendorarch}/Slurm
|
|
|
|
|
%{perl_vendorarch}/auto/Slurm
|
|
|
|
|
%{perl_vendorarch}/Slurmdb.pm
|
|
|
|
|
%{perl_vendorarch}/auto/Slurmdb
|
|
|
|
|
%{_mandir}/man3/Slurm*.3pm.*
|
|
|
|
|
|
|
|
|
|
%files slurmdbd
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_sbindir}/slurmdbd
|
|
|
|
|
%{_mandir}/man5/slurmdbd.*
|
|
|
|
|
%{_mandir}/man8/slurmdbd.*
|
|
|
|
|
%config(noreplace) %{_sysconfdir}/%{name}/slurmdbd.conf
|
Accepting request 441490 from home:eeich:branches:network:cluster
- Fix build with and without OHCP_BUILD define.
- Fix build for systemd and non-systemd.
- Updated to 16-05-5 - equvalent to OpenHPC 1.2.
* Fix issue with resizing jobs and limits not be kept track of correctly.
* BGQ - Remove redeclaration of job_read_lock.
* BGQ - Tighter locks around structures when nodes/cables change state.
* Make it possible to change CPUsPerTask with scontrol.
* Make it so scontrol update part qos= will take away a partition QOS from
a partition.
* Backfill scheduling properly synchronized with Cray Node Health Check.
Prior logic could result in highest priority job getting improperly
postponed.
* Make it so daemons also support TopologyParam=NoInAddrAny.
* If scancel is operating on large number of jobs and RPC responses from
slurmctld daemon are slow then introduce a delay in sending the cancel job
requests from scancel in order to reduce load on slurmctld.
* Remove redundant logic when updating a job's task count.
* MySQL - Fix querying jobs with reservations when the id's have rolled.
* Perl - Fix use of uninitialized variable in slurm_job_step_get_pids.
* Launch batch job requsting --reboot after the boot completes.
* Do not attempt to power down a node which has never responded if the
slurmctld daemon restarts without state.
* Fix for possible slurmstepd segfault on invalid user ID.
* MySQL - Fix for possible race condition when archiving multiple clusters
at the same time.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Remove the SchedulerParameters option of "assoc_limit_continue", making it
the default value. Add option of "assoc_limit_stop". If "assoc_limit_stop"
is set and a job cannot start due to association limits, then do not attempt
to initiate any lower priority jobs in that partition. Setting this can
decrease system throughput and utlization, but avoid potentially starving
larger jobs by preventing them from launching indefinitely.
* Update a node's socket and cores per socket counts as needed after a node
boot to reflect configuration changes which can occur on KNL processors.
Note that the node's total core count must not change, only the distribution
of cores across varying socket counts (KNL NUMA nodes treated as sockets by
Slurm).
* Rename partition configuration from "Shared" to "OverSubscribe". Rename
salloc, sbatch, srun option from "--shared" to "--oversubscribe". The old
options will continue to function. Output field names also changed in
scontrol, sinfo, squeue and sview.
* Add SLURM_UMASK environment variable to user job.
* knl_conf: Added new configuration parameter of CapmcPollFreq.
* Cleanup two minor Coverity warnings.
* Make it so the tres units in a job's formatted string are converted like
they are in a step.
* Correct partition's MaxCPUsPerNode enforcement when nodes are shared by
multiple partitions.
* node_feature/knl_cray - Prevent slurmctld GRES errors for "hbm" references.
* Display thread name instead of thread id and remove process name in stderr
logging for "thread_id" LogTimeFormat.
* Log IP address of bad incomming message to slurmctld.
* If a user requests tasks, nodes and ntasks-per-node and
tasks-per-node/nodes != tasks print warning and ignore ntasks-per-node.
* Release CPU "owner" file locks.
* Update seff to fix warnings with ncpus, and list slurm-perlapi dependency
in spec file.
* Allow QOS timelimit to override partition timelimit when EnforcePartLimits
is set to all/any.
* Make it so qsub will do a "basename" on a wrapped command for the output
and error files.
* Add logic so that slurmstepd can be launched under valgrind.
* Increase buffer size to read /proc/*/stat files.
* Prevent job stuck in configuring state if slurmctld daemon restarted while
PrologSlurmctld is running. Also re-issue burst_buffer/pre-load operation
as needed.
* Move test for job wait reason value of BurstBufferResources and
BurstBufferStageIn later in the scheduling logic.
* Document which srun options apply to only job, only step, or job and step
allocations.
* Use more compatible function to get thread name (>= 2.6.11).
* Make it so the extern step uses a reverse tree when cleaning up.
* If extern step doesn't get added into the proctrack plugin make sure the
sleep is killed.
* Add web links to Slurm Diamond Collectors (from Harvard University) and
collectd (from EDF).
* Add job_submit plugin for the "reboot" field.
* Make some more Slurm constants (INFINITE, NO_VAL64, etc.) available to
job_submit/lua plugins.
* Send in a -1 for a taskid into spank_task_post_fork for the extern_step.
* MYSQL - Sightly better logic if a job completion comes in with an end time
of 0.
* task/cgroup plugin is configured with ConstrainRAMSpace=yes, then set soft
memory limit to allocated memory limit (previously no soft limit was set).
* Streamline when schedule() is called when running with message aggregation
on batch script completes.
* Fix incorrect casting when [un]packing derived_ec on slurmdb_job_rec_t.
* Document that persistent burst buffers can not be created or destroyed using
the salloc or srun --bb options.
* Add support for setting the SLURM_JOB_ACCOUNT, SLURM_JOB_QOS and
SLURM_JOB_RESERVAION environment variables are set for the salloc command.
Document the same environment variables for the salloc, sbatch and srun
commands in their man pages.
* Fix issue where sacctmgr load cluster.cfg wouldn't load associations
that had a partition in them.
* Don't return the extern step from sstat by default.
* In sstat print 'extern' instead of 4294967295 for the extern step.
* Make advanced reservations work properly with core specialization.
* slurmstepd modified to pre-load all relevant plugins at startup to avoid
the possibility of modified plugins later resulting in inconsistent API
or data structures and a failure of slurmstepd.
* Export functions from parse_time.c in libslurm.so.
* Export unit convert functions from slurm_protocol_api.c in libslurm.so.
* Fix scancel to allow multiple steps from a job to be cancelled at once.
* Update and expand upgrade guide (in Quick Start Administrator web page).
* burst_buffer/cray: Requeue, but do not hold a job which fails the pre_run
operation.
* Insure reported expected job start time is not in the past for pending jobs.
* Add support for PMIx v2.
OBS-URL: https://build.opensuse.org/request/show/441490
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=12
2016-11-24 23:01:51 +01:00
|
|
|
%{_sysconfdir}/%{name}/slurmdbd.conf.example
|
|
|
|
|
%if 0%{?with_systemd}
|
2018-03-15 13:23:19 +01:00
|
|
|
%{_unitdir}/slurmdbd.service
|
2014-11-02 21:27:29 +01:00
|
|
|
%else
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_initrddir}/slurmdbd
|
2014-11-02 21:27:29 +01:00
|
|
|
%endif
|
2014-11-04 21:12:14 +01:00
|
|
|
%{_sbindir}/rcslurmdbd
|
2013-04-08 23:59:36 +02:00
|
|
|
|
2017-10-05 07:13:44 +02:00
|
|
|
%files sql
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2017-10-05 07:13:44 +02:00
|
|
|
%dir %{_libdir}/slurm
|
|
|
|
|
%{_libdir}/slurm/accounting_storage_mysql.so
|
|
|
|
|
%{_libdir}/slurm/jobcomp_mysql.so
|
|
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%files plugins
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2018-03-15 13:23:19 +01:00
|
|
|
%config %{_sysconfdir}/ld.so.conf.d/slurm.conf
|
2013-04-08 23:59:36 +02:00
|
|
|
%dir %{_libdir}/slurm
|
2018-03-15 08:03:02 +01:00
|
|
|
%{_libdir}/slurm/libslurmfull.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/accounting_storage_filetxt.so
|
|
|
|
|
%{_libdir}/slurm/accounting_storage_none.so
|
|
|
|
|
%{_libdir}/slurm/accounting_storage_slurmdbd.so
|
2015-11-01 17:14:37 +01:00
|
|
|
%{_libdir}/slurm/acct_gather_energy_cray.so
|
|
|
|
|
%{_libdir}/slurm/acct_gather_energy_ibmaem.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/acct_gather_energy_none.so
|
|
|
|
|
%{_libdir}/slurm/acct_gather_energy_rapl.so
|
Accepting request 226317 from home:scorot:branches:network:cluster
- update to version 2.6.7
* Support for job arrays, which increases performance and ease of
use for sets of similar jobs.
* Job profiling capability added to record a wide variety of job
characteristics for each task on a user configurable periodic
basis. Data currently available includes CPU use, memory use,
energy use, Infiniband network use, Lustre file system use, etc.
* Support for MPICH2 using PMI2 communications interface with much
greater scalability.
* Prolog and epilog support for advanced reservations.
* Much faster throughput for job step execution with --exclusive
option. The srun process is notified when resources become
available rather than periodic polling.
* Support improved for Intel MIC (Many Integrated Core) processor.
* Advanced reservations with hostname and core counts now supports
asymmetric reservations (e.g. specific different core count for
each node).
* External sensor plugin infrastructure added to record power
consumption, temperature, etc.
* Improved performance for high-throughput computing.
* MapReduce+ support (launches ~1000x faster, runs ~10x faster).
* Added "MaxCPUsPerNode" partition configuration parameter. This
can be especially useful to schedule GPUs. For example a node
can be associated with two Slurm partitions (e.g. "cpu" and
"gpu") and the partition/queue "cpu" could be limited to only a
subset of the node's CPUs, insuring that one or more CPUs would
be available to jobs in the "gpu" partition/queue.
OBS-URL: https://build.opensuse.org/request/show/226317
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=4
2014-03-16 21:42:08 +01:00
|
|
|
%{_libdir}/slurm/acct_gather_filesystem_lustre.so
|
|
|
|
|
%{_libdir}/slurm/acct_gather_filesystem_none.so
|
2018-03-15 08:03:02 +01:00
|
|
|
%{_libdir}/slurm/acct_gather_interconnect_none.so
|
Accepting request 226317 from home:scorot:branches:network:cluster
- update to version 2.6.7
* Support for job arrays, which increases performance and ease of
use for sets of similar jobs.
* Job profiling capability added to record a wide variety of job
characteristics for each task on a user configurable periodic
basis. Data currently available includes CPU use, memory use,
energy use, Infiniband network use, Lustre file system use, etc.
* Support for MPICH2 using PMI2 communications interface with much
greater scalability.
* Prolog and epilog support for advanced reservations.
* Much faster throughput for job step execution with --exclusive
option. The srun process is notified when resources become
available rather than periodic polling.
* Support improved for Intel MIC (Many Integrated Core) processor.
* Advanced reservations with hostname and core counts now supports
asymmetric reservations (e.g. specific different core count for
each node).
* External sensor plugin infrastructure added to record power
consumption, temperature, etc.
* Improved performance for high-throughput computing.
* MapReduce+ support (launches ~1000x faster, runs ~10x faster).
* Added "MaxCPUsPerNode" partition configuration parameter. This
can be especially useful to schedule GPUs. For example a node
can be associated with two Slurm partitions (e.g. "cpu" and
"gpu") and the partition/queue "cpu" could be limited to only a
subset of the node's CPUs, insuring that one or more CPUs would
be available to jobs in the "gpu" partition/queue.
OBS-URL: https://build.opensuse.org/request/show/226317
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=4
2014-03-16 21:42:08 +01:00
|
|
|
%{_libdir}/slurm/acct_gather_profile_none.so
|
2015-11-01 17:14:37 +01:00
|
|
|
%{_libdir}/slurm/burst_buffer_generic.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/checkpoint_none.so
|
|
|
|
|
%{_libdir}/slurm/checkpoint_ompi.so
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_libdir}/slurm/core_spec_cray.so
|
|
|
|
|
%{_libdir}/slurm/core_spec_none.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/crypto_openssl.so
|
Accepting request 226317 from home:scorot:branches:network:cluster
- update to version 2.6.7
* Support for job arrays, which increases performance and ease of
use for sets of similar jobs.
* Job profiling capability added to record a wide variety of job
characteristics for each task on a user configurable periodic
basis. Data currently available includes CPU use, memory use,
energy use, Infiniband network use, Lustre file system use, etc.
* Support for MPICH2 using PMI2 communications interface with much
greater scalability.
* Prolog and epilog support for advanced reservations.
* Much faster throughput for job step execution with --exclusive
option. The srun process is notified when resources become
available rather than periodic polling.
* Support improved for Intel MIC (Many Integrated Core) processor.
* Advanced reservations with hostname and core counts now supports
asymmetric reservations (e.g. specific different core count for
each node).
* External sensor plugin infrastructure added to record power
consumption, temperature, etc.
* Improved performance for high-throughput computing.
* MapReduce+ support (launches ~1000x faster, runs ~10x faster).
* Added "MaxCPUsPerNode" partition configuration parameter. This
can be especially useful to schedule GPUs. For example a node
can be associated with two Slurm partitions (e.g. "cpu" and
"gpu") and the partition/queue "cpu" could be limited to only a
subset of the node's CPUs, insuring that one or more CPUs would
be available to jobs in the "gpu" partition/queue.
OBS-URL: https://build.opensuse.org/request/show/226317
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=4
2014-03-16 21:42:08 +01:00
|
|
|
%{_libdir}/slurm/ext_sensors_none.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/gres_gpu.so
|
|
|
|
|
%{_libdir}/slurm/gres_mic.so
|
|
|
|
|
%{_libdir}/slurm/gres_nic.so
|
|
|
|
|
%{_libdir}/slurm/jobacct_gather_cgroup.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/jobacct_gather_linux.so
|
|
|
|
|
%{_libdir}/slurm/jobacct_gather_none.so
|
|
|
|
|
%{_libdir}/slurm/jobcomp_filetxt.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/jobcomp_none.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/jobcomp_script.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/job_container_cncu.so
|
|
|
|
|
%{_libdir}/slurm/job_container_none.so
|
|
|
|
|
%{_libdir}/slurm/job_submit_all_partitions.so
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_libdir}/slurm/job_submit_cray.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/job_submit_defaults.so
|
|
|
|
|
%{_libdir}/slurm/job_submit_logging.so
|
|
|
|
|
%{_libdir}/slurm/job_submit_partition.so
|
Accepting request 226317 from home:scorot:branches:network:cluster
- update to version 2.6.7
* Support for job arrays, which increases performance and ease of
use for sets of similar jobs.
* Job profiling capability added to record a wide variety of job
characteristics for each task on a user configurable periodic
basis. Data currently available includes CPU use, memory use,
energy use, Infiniband network use, Lustre file system use, etc.
* Support for MPICH2 using PMI2 communications interface with much
greater scalability.
* Prolog and epilog support for advanced reservations.
* Much faster throughput for job step execution with --exclusive
option. The srun process is notified when resources become
available rather than periodic polling.
* Support improved for Intel MIC (Many Integrated Core) processor.
* Advanced reservations with hostname and core counts now supports
asymmetric reservations (e.g. specific different core count for
each node).
* External sensor plugin infrastructure added to record power
consumption, temperature, etc.
* Improved performance for high-throughput computing.
* MapReduce+ support (launches ~1000x faster, runs ~10x faster).
* Added "MaxCPUsPerNode" partition configuration parameter. This
can be especially useful to schedule GPUs. For example a node
can be associated with two Slurm partitions (e.g. "cpu" and
"gpu") and the partition/queue "cpu" could be limited to only a
subset of the node's CPUs, insuring that one or more CPUs would
be available to jobs in the "gpu" partition/queue.
OBS-URL: https://build.opensuse.org/request/show/226317
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=4
2014-03-16 21:42:08 +01:00
|
|
|
%{_libdir}/slurm/job_submit_require_timelimit.so
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_libdir}/slurm/job_submit_throttle.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/launch_slurm.so
|
2015-11-01 17:14:37 +01:00
|
|
|
%{_libdir}/slurm/layouts_power_cpufreq.so
|
|
|
|
|
%{_libdir}/slurm/layouts_power_default.so
|
|
|
|
|
%{_libdir}/slurm/layouts_unit_default.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/mcs_account.so
|
|
|
|
|
%{_libdir}/slurm/mcs_group.so
|
|
|
|
|
%{_libdir}/slurm/mcs_none.so
|
|
|
|
|
%{_libdir}/slurm/mcs_user.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/mpi_none.so
|
|
|
|
|
%{_libdir}/slurm/mpi_openmpi.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/mpi_pmi2.so
|
2015-11-01 17:14:37 +01:00
|
|
|
%{_libdir}/slurm/power_none.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/preempt_none.so
|
|
|
|
|
%{_libdir}/slurm/preempt_partition_prio.so
|
|
|
|
|
%{_libdir}/slurm/preempt_qos.so
|
|
|
|
|
%{_libdir}/slurm/priority_basic.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/priority_multifactor.so
|
|
|
|
|
%{_libdir}/slurm/proctrack_cgroup.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/proctrack_linuxproc.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/proctrack_pgid.so
|
2015-08-06 21:51:59 +02:00
|
|
|
%{_libdir}/slurm/route_default.so
|
|
|
|
|
%{_libdir}/slurm/route_topology.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/sched_backfill.so
|
|
|
|
|
%{_libdir}/slurm/sched_builtin.so
|
|
|
|
|
%{_libdir}/slurm/sched_hold.so
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_libdir}/slurm/select_alps.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/select_cons_res.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/select_cray.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/select_linear.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/select_serial.so
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_libdir}/slurm/slurmctld_nonstop.so
|
|
|
|
|
%{_libdir}/slurm/switch_cray.so
|
|
|
|
|
%{_libdir}/slurm/switch_generic.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/switch_none.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/task_affinity.so
|
|
|
|
|
%{_libdir}/slurm/task_cgroup.so
|
2014-07-26 16:36:03 +02:00
|
|
|
%{_libdir}/slurm/task_cray.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/task_none.so
|
|
|
|
|
%{_libdir}/slurm/topology_3d_torus.so
|
2015-11-01 17:14:37 +01:00
|
|
|
%{_libdir}/slurm/topology_hypercube.so
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_libdir}/slurm/topology_node_rank.so
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_libdir}/slurm/topology_none.so
|
|
|
|
|
%{_libdir}/slurm/topology_tree.so
|
2017-02-02 21:23:02 +01:00
|
|
|
%if 0%{?suse_version} > 1310
|
2018-03-15 08:03:02 +01:00
|
|
|
%{_libdir}/slurm/acct_gather_interconnect_ofed.so
|
2017-02-02 21:23:02 +01:00
|
|
|
%endif
|
|
|
|
|
%if 0%{?suse_version} > 1140
|
|
|
|
|
%ifarch %{ix86} x86_64
|
|
|
|
|
%{_libdir}/slurm/acct_gather_energy_ipmi.so
|
2019-01-31 21:14:27 +01:00
|
|
|
%{_libdir}/slurm/acct_gather_energy_xcc.so
|
2017-02-02 21:23:02 +01:00
|
|
|
%endif
|
|
|
|
|
%endif
|
|
|
|
|
%{_libdir}/slurm/node_features_knl_generic.so
|
|
|
|
|
|
|
|
|
|
%files lua
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2017-02-02 21:23:02 +01:00
|
|
|
%{_libdir}/slurm/job_submit_lua.so
|
|
|
|
|
%{_libdir}/slurm/proctrack_lua.so
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%files torque
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/pbsnodes
|
Accepting request 226317 from home:scorot:branches:network:cluster
- update to version 2.6.7
* Support for job arrays, which increases performance and ease of
use for sets of similar jobs.
* Job profiling capability added to record a wide variety of job
characteristics for each task on a user configurable periodic
basis. Data currently available includes CPU use, memory use,
energy use, Infiniband network use, Lustre file system use, etc.
* Support for MPICH2 using PMI2 communications interface with much
greater scalability.
* Prolog and epilog support for advanced reservations.
* Much faster throughput for job step execution with --exclusive
option. The srun process is notified when resources become
available rather than periodic polling.
* Support improved for Intel MIC (Many Integrated Core) processor.
* Advanced reservations with hostname and core counts now supports
asymmetric reservations (e.g. specific different core count for
each node).
* External sensor plugin infrastructure added to record power
consumption, temperature, etc.
* Improved performance for high-throughput computing.
* MapReduce+ support (launches ~1000x faster, runs ~10x faster).
* Added "MaxCPUsPerNode" partition configuration parameter. This
can be especially useful to schedule GPUs. For example a node
can be associated with two Slurm partitions (e.g. "cpu" and
"gpu") and the partition/queue "cpu" could be limited to only a
subset of the node's CPUs, insuring that one or more CPUs would
be available to jobs in the "gpu" partition/queue.
OBS-URL: https://build.opensuse.org/request/show/226317
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=4
2014-03-16 21:42:08 +01:00
|
|
|
%{_bindir}/qalter
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/qdel
|
|
|
|
|
%{_bindir}/qhold
|
|
|
|
|
%{_bindir}/qrls
|
Accepting request 226317 from home:scorot:branches:network:cluster
- update to version 2.6.7
* Support for job arrays, which increases performance and ease of
use for sets of similar jobs.
* Job profiling capability added to record a wide variety of job
characteristics for each task on a user configurable periodic
basis. Data currently available includes CPU use, memory use,
energy use, Infiniband network use, Lustre file system use, etc.
* Support for MPICH2 using PMI2 communications interface with much
greater scalability.
* Prolog and epilog support for advanced reservations.
* Much faster throughput for job step execution with --exclusive
option. The srun process is notified when resources become
available rather than periodic polling.
* Support improved for Intel MIC (Many Integrated Core) processor.
* Advanced reservations with hostname and core counts now supports
asymmetric reservations (e.g. specific different core count for
each node).
* External sensor plugin infrastructure added to record power
consumption, temperature, etc.
* Improved performance for high-throughput computing.
* MapReduce+ support (launches ~1000x faster, runs ~10x faster).
* Added "MaxCPUsPerNode" partition configuration parameter. This
can be especially useful to schedule GPUs. For example a node
can be associated with two Slurm partitions (e.g. "cpu" and
"gpu") and the partition/queue "cpu" could be limited to only a
subset of the node's CPUs, insuring that one or more CPUs would
be available to jobs in the "gpu" partition/queue.
OBS-URL: https://build.opensuse.org/request/show/226317
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=4
2014-03-16 21:42:08 +01:00
|
|
|
%{_bindir}/qrerun
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/qstat
|
|
|
|
|
%{_bindir}/qsub
|
2017-05-30 12:41:04 +02:00
|
|
|
%{_bindir}/mpiexec.slurm
|
2017-10-05 07:13:44 +02:00
|
|
|
%{_bindir}/generate_pbs_nodefile
|
|
|
|
|
%{_libdir}/slurm/job_submit_pbs.so
|
|
|
|
|
%{_libdir}/slurm/spank_pbs.so
|
2013-04-08 23:59:36 +02:00
|
|
|
|
|
|
|
|
%files sjstat
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2013-04-08 23:59:36 +02:00
|
|
|
%{_bindir}/sjstat
|
|
|
|
|
|
|
|
|
|
%files pam_slurm
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2018-04-12 19:22:25 +02:00
|
|
|
%doc ../README.pam_slurm ../README.pam_slurm_adopt
|
2013-04-08 23:59:36 +02:00
|
|
|
/%_lib/security/pam_slurm.so
|
2015-11-01 17:14:37 +01:00
|
|
|
/%_lib/security/pam_slurm_adopt.so
|
|
|
|
|
|
2018-01-30 17:25:18 +01:00
|
|
|
%files node
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2018-01-30 17:25:18 +01:00
|
|
|
%{_sbindir}/slurmd
|
|
|
|
|
%{_sbindir}/slurmstepd
|
|
|
|
|
%{_mandir}/man8/slurmd.*
|
|
|
|
|
%{_mandir}/man8/slurmstepd*
|
|
|
|
|
%if 0%{?with_systemd}
|
|
|
|
|
%{_sbindir}/rcslurmd
|
|
|
|
|
%{_unitdir}/slurmd.service
|
|
|
|
|
%else
|
|
|
|
|
%{_initrddir}/slurm
|
|
|
|
|
%{_sbindir}/rcslurm
|
|
|
|
|
%endif
|
|
|
|
|
|
|
|
|
|
%files config
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2018-01-30 17:25:18 +01:00
|
|
|
%dir %{_sysconfdir}/%{name}
|
|
|
|
|
%dir %{_sysconfdir}/%{name}/layouts.d
|
|
|
|
|
%config(noreplace) %{_sysconfdir}/%{name}/slurm.conf
|
|
|
|
|
%{?OHPC_BUILD:%config %{_sysconfdir}/%{name}/slurm.conf.example}
|
|
|
|
|
%config(noreplace) %{_sysconfdir}/%{name}/cgroup.conf
|
|
|
|
|
%config(noreplace) %{_sysconfdir}/%{name}/layouts.d/power.conf.example
|
|
|
|
|
%config(noreplace) %{_sysconfdir}/%{name}/layouts.d/power_cpufreq.conf.example
|
|
|
|
|
%config(noreplace) %{_sysconfdir}/%{name}/layouts.d/unit.conf.example
|
|
|
|
|
%{?OHPC_BUILD:%attr(0755, %slurm_u, %slurm_g) %_localstatedir/lib/slurm}
|
|
|
|
|
%{?with_systemd:%{_tmpfilesdir}/%{name}.conf}
|
|
|
|
|
%dir %{_var}/spool/slurm
|
2018-12-12 10:28:26 +01:00
|
|
|
%config(noreplace) %{_sysconfdir}/logrotate.d/slurm*
|
2018-11-20 18:07:44 +01:00
|
|
|
|
|
|
|
|
%files config-man
|
2019-02-01 20:44:10 +01:00
|
|
|
%{?comp_at}
|
2018-09-30 17:33:20 +02:00
|
|
|
%{_mandir}/man5/acct_gather.conf.*
|
|
|
|
|
%{_mandir}/man5/burst_buffer.conf.*
|
|
|
|
|
%{_mandir}/man5/ext_sensors.conf.*
|
|
|
|
|
%{_mandir}/man5/slurm.*
|
|
|
|
|
%{_mandir}/man5/cgroup.*
|
|
|
|
|
%{_mandir}/man5/cray.*
|
|
|
|
|
%{_mandir}/man5/gres.*
|
|
|
|
|
%{_mandir}/man5/nonstop.conf.5.*
|
|
|
|
|
%{_mandir}/man5/topology.*
|
|
|
|
|
%{_mandir}/man5/knl.conf.5.*
|
2018-01-30 17:25:18 +01:00
|
|
|
|
2013-04-08 23:59:36 +02:00
|
|
|
%changelog
|
