------------------------------------------------------------------- Thu Jul 25 02:17:50 UTC 2013 - xli@suse.com - Fix corosync start failed issue ------------------------------------------------------------------- Tue Jul 23 09:44:07 UTC 2013 - xli@suse.com - Update to corosync 2.3.1 stable release ------------------------------------------------------------------- Fri May 10 13:18:25 UTC 2013 - tserong@suse.com - Update to corosync 1.4.5 stable release (bnc#799031) - coroipc: Handle pfd.revents as bit-field - Check socket_recv error code in ipc_dispatch_get - On places with POLLERR check also POLLNVAL - coroipc: Don't spin when waiting on semaphore - log: Handle race in printf_to_logs and format_set - objdb: Don't read uninitialized memory in inc/dec - Add waiting_trans_ack also to fragmentation layer - Handle segfault in backlog_get - Fix problem with sync operations under very rare circumstances - manpages: Add confdb_key_get man page - manpages: Add links for referenced confdb calls - manpages: Fix typo in evs* manpages - If failed_to_recv is set, consensus can be empty - Ignore sync barrier msgs if sync doesn't started - Make service_build contain correct number of msgs - Handle sync and service unload correctly - Don't call sync_* funcs for unloaded services - Return back "Totem is unable to form..." message - Move "Totem is unable to form..." message to main - Use unix socket for local multicast loop - cpg: Enhance downlist selection algorithm - cpg: Process join list after downlists - cpg: Never choose downlist with localnode - Fix cpg_membership_get() - Don't access invalid mem in totemconfig - Move some totem and cpg messages to trace level - flatiron: Free outq items list on conn exit - Fix nodeid conflicting issue (bnc#806634) + Added fix-nodeid-conflicting.patch - change the default priority to RR(1) same as pacemaker(bnc#804707) + Added corosync_reduce_RR_priority.patch ------------------------------------------------------------------- Thu Mar 21 11:59:58 UTC 2013 - mmeister@suse.com - Added url as source. Please see http://en.opensuse.org/SourceUrls ------------------------------------------------------------------- Fri Jun 8 07:46:10 UTC 2012 - tserong@suse.com - Update to corosync 1.4.3 stable release. - Add calls to missing object_find_destroy() to fix mem leaks - Free mem allocated by getaddrinfo - corosync.conf.example: change bindnetaddr, mcastaddr, add comments - Store error str if can't open logfile - Wait for corosync-notifyd exit in init script - iba: Use configured node id - Unlink shm buffers if init fails - Fix memory leaks when nss fails - Madvise NOSYNC flag only if available - Include net/if_var.h header only when needed - Include stdint.h because funcs uses int16_t - Use install instead of cp - Don't unlock mutex in different threads - Revert "Use install instead of cp" - Add support for per OS CP flags - Remove cloned lines in main of main.c - Fixed bug when corosync receive JoinMSG in OPERATIONAL state - Correct nodeid of token when we retransmit it - Correct nodeid in memb_state_commit_token_send function ------------------------------------------------------------------- Sun Feb 5 11:44:40 UTC 2012 - jjzhang@suse.com - Send CPG_REASON_PROCDOWN when really needed (bnc#740343) ------------------------------------------------------------------- Tue Sep 20 13:15:22 UTC 2011 - tserong@suse.com - Changes since corosync 1.4.1 stable release: * Resolve a deadlock between the timer and serialize locks. * totemconfig: change minimum RRP threshold * Ignore memb_join messages during flush operations * rrp: Higher threshold in passive mode for mcast (bnc#712037) * rrp: Handle endless loop if all ifaces are faulty (bnc#712037) * A CPG client can sometimes lockup if the local node is in the downlist * Handle errors from totem_mcast * coroipcc: use malloc for path in service_connect * Version cs: 23112099e1c2b620e6976ca099d2b9afc80721aa - corosync 1.4.1 stable release: * main: let poll really stop before totempg_finalize * totemsrp: fix buffer overflows for large clusters (> 100 nodes) * rrp: Handle rollower in passive rrp properly * rrp: handle rollover in active rrp properly * totemconfig: Change default FAIL_TO_RECV_CONST * Fix problem where corosync will segfault if there are gaps in recovery queue * cpgtool/cfgtool: print list of IP with space between items * RRP: redundant ring automatic recovery (fate#310284) * fix typos in cpg_mcast_joined.3 and cpg_zcb_mcast_joined.3 * Remove spinlocks * confdb: Resolve dispatch deadlock * RRP: Fix ring initialization issue for UDPU mode * crypto: rng_make_prng prevent buf overflow * cpg: do_proc_join change list_slice to list_add * totemudp: memset of proper size * coroipcs: init buf in coroipcs_handler_dispatch * iazc: Reduce number of mem alloc and memcpy * coroipcc: Fix unhandled BSD EOF in coroipcc_dispatch_get() * cpg: fix sync master selection when one node paused * totemsrp: Enhance mcast failure detection * coroipcs: Deny connect to service without initfn * Add ipc_refcnt to message_handler_req_{exec, lib}_cfg_ringreenable() - corosync 1.3.1 release: * corosync crashing when a network becomes disrupted and then restored (bnc#685241) * Align IPC on 8 byte boundaries for performance and avoid bus errors. * Provide better checking of the message type. * totemsrp: free messages originated in recovery rather then rely on messages_free * Resolve abort during simulatenous stopping of at least 4 nodes. * Don't assert when ring id file is less then 8 bytes (possibly after local fs problems). * Handle delayed multicast packets that occur with switches. * CPG: make sure coroipcc_service_disconnect() is always called. * Fix abort when token is lost in RECOVERY state (bnc#677779) ------------------------------------------------------------------- Sat Sep 17 23:53:47 UTC 2011 - jengelh@medozas.de - Remove redundant tags/sections from specfile - Add baselibs configuration ------------------------------------------------------------------- Tue Feb 8 13:03:11 UTC 2011 - tserong@novell.com - Update to corosync 1.3.0 - Set the max buffer size for sockets to reduce message dropping - diags: add a mechanism to trigger the writing the flight data - Add the UDPU transport (UDP transport for corosync) - Remove delay in library on corosync shutdown - Check for a properly configured multicast address. - cpg: fix sync'ing the downlist. - POLL: gracefully handle running out of file descriptors. - Return CS_ERR_NO_RESOURCES when the server is low on available file descriptors. - Remove checking of subparameters in service.d files. - Only allow corosync to run one copy via a lock file. - When used with the openais ckpt service, don't disconnect an ipc connection during configuration change that takes longer then 2 seconds. - Remove the token cancel retransmit timeout on receipt of a multicast message. ------------------------------------------------------------------- Thu Aug 5 04:55:08 UTC 2010 - tserong@novell.com - Update to corosync 1.2.7 - Remove consensus check for two node cluster cases which can have smaller consensus values. Document in man page the behavior of consensus. ------------------------------------------------------------------- Tue Jul 27 11:48:21 UTC 2010 - tserong@novell.com - Fix problem where flow control could lock up ipc under very heavy load in very rare circumstances (upstream cs 3003) - SYNC: always call sync_aborted() in sync_confchg_fn() (upstream cs 3000) - SYNCV2: reset the my_memb_determine_ring_id in sync_v2_memb_list_abort() (upstream cs 2999) - Fix logging_daemon config parser code (rhbz#615203) (upstream cs 2998) - Remove reset of token timeout on retransmitted token reception. Fixes membership problems with certain timing parametrs (upstream cs 2989) - Speed up IPC connection process (upstream cs 2987) - Fix fail list fault that occurs in very rare circumstances (upstream cs 2985) ------------------------------------------------------------------- Thu Jul 22 03:31:59 UTC 2010 - tserong@novell.com - Update to corosync 1.2.6 - 80% packet loss networks were resulting in problems with totem. - Fixed ~40 scanning errors found with coverity. - cpg_membership_get now functional. - errors logged prior to the start of the daemon were not flushed. - Fixes defects in logsys which are crashing pacemaker installations. - Adds man pages for all binaries - Fixes several defects found in high packet loss field environments. - Send proper notification code of CPG_REASON_LEAVE in cpg service. - Fix segfault when pacemaker forks new processes - Unlock global serializer lock during shutdown to prevent spinning on single cpu systems or high cpu use on mulitple cpu systems - Stop totem statistics updater timer during shutdown to prevent a segfault during shutdown. - Fix problem where glibc's fork() implementation may cause segfaults in Pacemaker's use of the fork() system call. - Fix problem where a full /dev/shm would result in client segfault - instead an error is returned in this situation. - Fix problem where flight recorder leaks files in shared memory filesystem. Also clean up the error handling of the shared memory allocation code of the flight recorder. - Fix problem where a failure in glibc's pathconf API would result in segfault. - Add corosync and corosync-blackbox man pages. ------------------------------------------------------------------- Fri Jul 9 08:53:55 UTC 2010 - dmuhamedagic@novell.com - prevent corosync-cfgtool from hanging (bnc#616183) ------------------------------------------------------------------- Wed Jun 2 11:53:28 UTC 2010 - tserong@novell.com - Set sensible defaults for Pacemaker in corosync.conf.example (bnc#610663) - Clarify bindnetaddr option in corosync.conf.5 manpage (upstream cs 2856) ------------------------------------------------------------------- Mon May 10 14:59:13 UTC 2010 - dmuhamedagic@novell.com - Handle POLLNVAL in coroipcc - Save the ring id and restore it properly when the recovery operation fails - increase maximum entries in the retransmit queue when recovery takes place. - fix one-off error in memove - discard and report unknown messages ------------------------------------------------------------------- Mon Apr 26 14:40:44 UTC 2010 - dmuhamedagic@novell.com - fix valgrind reported problems (upstream cs 2787) - Memset for res_setup variable in coroipcs:req_setup_send - Two memset in logsys for buffers - Problem in corosync_totem_stats_updater where avg_token_holdtime has size of avg_backlog_calc - corosync_totem_stats_init where avg_backlog_calc is 32 bits (not 64) - objdb problem if new_valie_len != object->value_len. In such case newly allocated memory is not initialized and in some situations, value_len is not updated. - select a new sync member if the node with the lowest nodeid has left (upstream cs 2785) - fix a crash in YKD ------------------------------------------------------------------- Fri Apr 9 15:09:11 UTC 2010 - dmuhamedagic@novell.com - clear the ring id on sync abort (bnc#590666) - fix unloading of evs - change sign of all exit codes (normal error exit is now 1) - objdb: fix key change notifications (don't notify if the key wasn't changed; notify on key inc/dec) - fix possible lockup when a dispatch handler function is NULL - upstream version cs 2756 ------------------------------------------------------------------- Mon Mar 29 14:15:43 UTC 2010 - dmuhamedagic@novell.com - fix lockup that occurs sometimes before exiting - fix problem where retransmissions don't occur resulting in failure to receive condition - add a reload callback to libconfdb - support for lib_cpg_finalize - cpg join with undelivered leave message (fixes problems with nodes joining cpg twice in quick succession) - fix error handling to avoid segfaults/leaks on error in coroipcc_service_connect - upstream release 1.2.1 ------------------------------------------------------------------- Thu Mar 4 18:43:07 UTC 2010 - dmuhamedagic@novell.com - retain nodeid compatibility with openais (revert patch from cs 2429) ------------------------------------------------------------------- Wed Mar 3 16:41:12 CST 2010 - jjzhang@novell.com - minor enhancement to corosync.conf man page (bnc#580180) ------------------------------------------------------------------- Tue Mar 2 22:01:26 UTC 2010 - dmuhamedagic@novell.com - upstream version cs 2667 - allow empty (default) consensus timeout - fix freeze of IPC library connection on sem_wait - fix malloc deadlock in signal handler (rhbz#547511) - fix coroipcs message corruption that occurs when a message fills the remainder of the dispatch buffer with a full message - totemsrp: fix transitional configuration changes with long token timeouts - remove a double list_del() when a tracking CFG client shuts down without calling cfg_track_stop (it caused corosync to crash) - use nodeid instead of localhost ip for the case when binding to a loalhost interface - fix corosync shutdown process - add augeas lense for corosync.conf - patch to set unset value in token hold cancel structure as to not crash wireshark - convert unsafe function to thread-safe reentrant equivalents ------------------------------------------------------------------- Mon Feb 22 15:53:00 UTC 2010 - dmuhamedagic@novell.com - SP1 beta5 (no code changes) ------------------------------------------------------------------- Mon Feb 8 14:53:31 UTC 2010 - dmuhamedagic@novell.com - turn timestamp off in corosync.conf.example (there was a problem reported in connection with not thread-safe glibc functions used in concert with this option, which hasn't yet been resolved) ------------------------------------------------------------------- Wed Jan 27 13:44:58 UTC 2010 - dmuhamedagic@novell.com - add cs2646 patch from upstream, fixes cs2642 - add patch to accept on/off for the various log directives (bnc#573451) ------------------------------------------------------------------- Thu Jan 21 14:21:44 UTC 2010 - dmuhamedagic@novell.com - %pre script moved to openais ------------------------------------------------------------------- Mon Jan 18 16:36:24 UTC 2010 - dmuhamedagic@novell.com - remove init script - add %pre script to copy openais.conf and authkey to /etc/corosync - add patch 2642 (parser fix) - fix some obsoletes/requires ------------------------------------------------------------------- Mon Jan 11 16:08:35 UTC 2010 - lmb@novell.com - Update to corosync 1.2.0. ------------------------------------------------------------------- Tue Dec 29 10:23:21 UTC 2009 - dmuhamedagic@novell.com - init script changes: + replace killall with checkproc, otherwise corosync can't stop + test if sbd/lrmadmin exist, because corosync has no dependency on cluster-glue ------------------------------------------------------------------- Tue Dec 15 15:27:37 UTC 2009 - dmuhamedagic@novell.com - update to the corosync upstream release 1.2.0 - add suse init script - don't create rccorosync, because users should be using rcopenais to start a cluster - rename corosynclib to libcorosync4 (similar for the devel package) (http://en.opensuse.org/Shared_Library_Packaging_Policy) - Autotools generated version from the released upstream version 1.2.0 - some specfile changes (initddir -> initrddir, header)