Gitweb:
http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=...
Commit: 8f34db0d3dae25c9733cdff1f5df1af404280f3a
Parent: 0000000000000000000000000000000000000000
Author: Digimer <digimer(a)alteeve.ca>
AuthorDate: 2012-06-01 03:12 +0000
Committer: Digimer <digimer(a)alteeve.ca>
CommitterDate: 2012-06-01 03:12 +0000
annotated tag: cluster-3.1.91 has been created
at 8f34db0d3dae25c9733cdff1f5df1af404280f3a (tag)
tagging c0afc6f3180dd61c9b212e2eb61cdc8d9c56af9f (commit)
replaces cluster-3.1.90
cluster-3.1.91 release
David Teigland (2):
fenced: fix handling of startup partition merge
dlm_controld: fix handling of startup partition merge
Fabio M. Di Nitto (25):
test commit
Revert "test commit"
test commit
Revert "test commit"
qdiskd: small tiny little cleanup
cman: fix bad merge from coverity branch
config: update relax ng schema to include totem miss_count_const
cman init: fix start sequence error handling
man: update fenced.8 to reflect a limitation in XML/DTD implementation
cpglockd: cleanup startup init scripts and drop code from rgmanager
cpglockd: fix pid file creation and check if daemon is already running
cpglockd: fix a startup race condition
rgmanager: cleanup binary linking and include dirs
cpglock: drop unused define
cpglockd: make network comm endian free
cpglockd: use standard logging infrastructure
cpglockd: refuse to shutdown if clients are connected
rgmanager: die hard if we cpglockd dies
cpglockd: show that we are alive while waiting for quorum (debug info)
rgmanager: fix cpglockd vs rgmanager startup race condition with systemd
rgmanager: fix conditional check for cpglockd startup
qdisk: Fix man page example (take 2)
cmannotifyd: deliver cluster status at startup and fix daemon init
cman init: allow sysconfig/cman to pass options to dlm_controld
cman init: add extra documentation for FENCE_JOIN=""
Jan Pokorný (6):
resrules: fix free while passing the pointer to caller
resrules: print_resource_rule doc comment conformance
rgmanager: resrules: simplify getting attributes + avoid memleaks
rgmanager: reslist: nothing avoids using size_t
rgmanager: reslist: sanitize act_dup
rgmanager: reslist: another string related cleanup
John Ruemker (1):
cman: Improve quorum timer handling how quorum timers work
Lon Hohberger (17):
rgmanager: OCF_NOT_INSTALLED is OK if we stop when stopped
rgmanager: Handle REASON_TRY_SHUTDOWN in two ways
rgmanager: Add extra service status call
rgmanager: Avoid duplicate restart of service
rgmanager: Fix call to service_status()
rgmanager: Fix tiny memory leak in sl_service_status
rgmanager: Fix dependency issue related to ordering
Revert "rgmanager: Fix dependency issue related to ordering"
rgmanager: Fix dependency issue related to ordering
config: Fix typo in schema
cman: Allocate port in cnxman-socket.h for qdiskd
rgmanager: Retry when config is out of sync
rgmanager: Add simple locking over CPG
rgmanager: Merge upstream cpglockd
rgmanager: Transmogrify states to expected values
rgmanager: Retry connection to cpglockd
rgmanager: Add man page for cpglockd; update rgmanager.8
Marc Grimme (1):
rgmanager: Small bug in follow-service.sl
Ryan McCabe (31):
rgmanager: Sync cpglock bits with upstream
rgmanager: Sync cpglock bits with upstream
rgmanager: Start cpglockd automatically when RRP mode is enabled
rgmanager: Add a utility to dump the cpglockd state
Allow -C to take a 1 or 0 argument to respectively force enable
Cleanup and exit if we receive a SIGTERM
Add a simple init script for cpglockd. The rgmanager init script still
Update Makefile for cpglockd init script
Cleanup the Makefile and add support to the cpglockd init script for
Clean up the pid file when exiting after running with -f
Avoid a free() of an uninitialized variable when central
Convert all remaining calls to printf and perror for error reporting
If we see a JOIN message from ourselves, move ourselves to the end of
Fix operation when quorum is lost:
Print cpg_mcast_joined() return value and errno if it fails
Add some additional debugging logging for try-locks
Set m.owner_nodeid to the our node id when NAKing requests when
Copy lock messages before swapping bytes because we
Work around a fenced hang/delay that can result in us unpausing too soon
Poll for fenced node updates less aggressively while waiting
Replace printf and fprintf calls with logt_print in the
Increment the recovered locks count when a client is disconnected.
rgmanager: Install the new cpglockd man page
Apply fixes from John Ruemker to fix some obscure error cases when
rgmanager: Fix the function prototypes for act_dup() which were broken
rgmanager: Clean up lock handling in groups.c
rgmanager: Fix return with lock held
rgmanager: Fix dependent service handling during recovery
cpglockd: Fix a hang triggered when fencing completes in less than 1s
rgmanager: Fix a crash when dbus notifications are enabled
rgmanager: Revert patch that caused a regression when operating on resources
jpokorny(a)redhat.com (1):
rgmanager: various cleanups regarding string functions