- 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
- version 2.5.7
* Fix for linking to the select/cray plugin to not give warning
about undefined variable.
* Add missing symbols to the xlator.h
* Avoid placing pending jobs in AdminHold state due to backfill
scheduler interactions with advanced reservation.
* Accounting - make average by task not cpu.
* POE - Correct logic to support poe option "-euidevice sn_all"
and "-euidevice sn_single".
* Accounting - Fix minor initialization error.
* POE - Correct logic to support srun network instances count
with POE.
* POE - With the srun --launch-cmd option, report proper task
count when the --cpus-per-task option is used without the
--ntasks option.
* POE - Fix logic binding tasks to CPUs.
* sview - Fix race condition where new information could of
slipped past the node tab and we didn't notice.
* Accounting - Fix an invalid memory read when slurmctld sends
data about start job to slurmdbd.
* If a prolog or epilog failure occurs, drain the node rather
than setting it down and killing all of its jobs.
* Priority/multifactor - Avoid underflow in half-life calculation.
* POE - pack missing variable to allow fanout (more than 32
nodes)
* Prevent clearing reason field for pending jobs. This bug was
introduced in v2.5.5 (see "Reject job at submit time ...").
* BGQ - Fix issue with preemption on sub-block jobs where a job
would kill all preemptable jobs on the midplane instead of just
the ones it needed to.
OBS-URL: https://build.opensuse.org/request/show/177944
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=3
- version 2.5.4
* Support for Intel® Many Integrated Core (MIC) processors.
* User control over CPU frequency of each job step.
* Recording power usage information for each job.
* Advanced reservation of cores rather than whole nodes.
* Integration with IBM's Parallel Environment including POE (Parallel Operating Environment) and NRT (Network Resource Table) API.
* Highly optimized throughput for serial jobs in a new "select/serial" plugin.
* CPU load is information available
* Configurable number of CPUs available to jobs in each SLURM partition, which provides a mechanism to reserve CPUs for use with GPUs.
OBS-URL: https://build.opensuse.org/request/show/163479
OBS-URL: https://build.opensuse.org/package/show/network:cluster/slurm?expand=0&rev=2
