Page MenuHomeClusterLabs Projects
Files F4833503

ChangeLog
No OneTemporary
Actions

ChangeLog
View Options

2006-08-08 Lon Hohberger <lhh at redhat.com>
* src/clulib/members.c: Fix gained/lost list creation so that the
count is actually nonzero (#201713)
* src/daemons/main.c: Fix typo in debug information. Kill CMAN handle
before doing a member list update. Don't call memb_online; just
check the cn_member field (optimization). Don't process node event
until the node actually closes port (#201713).
* src/utils/clusvcadm.c: Fix relocation issues (#201711).
2006-08-07 Lon Hohberger <lhh at redhat.com>
* src/clulib/ckpt_state.c: Preliminary implementation of replacement
for VF using AIS CKPT B.02.01 (w/ built-in test program)
* include/cman-private.h: Clean up APIs (cman APIs return
cman_handle_t, which is void*, should be using void ** all over)
* include/message.h: Bump context count to 128, add destination
node ID in header of packets.
* src/clulib/alloc.c: If we alloc the same size, return the same
block
* src/clulib/cman.c: API cleanups
* src/clulib/message.c: Add error checking to msg_print
* src/clulib/msg_cluster.c: Check destination in header before
processing message remove dup #define for MAX_CONTEXTS, add
proto_error() macro for displaying protocol errors. Use 'max'
instead of 'fd' for select(). Use correct var when assigning
contexts. Fix CMAN handles. Return correct size from msg_send()
requests.
* src/clulib/msgtest.c: Fix CMAN handles
* src/clulib/vft.c: Don't handle VF_CURRENT inside comms thread
* src/daemons/main.c: Check to see if nodes are listening on our
port before we consider them running. Handle VF_CURRENT requests
from other nodes. Fail if we can't determine local node ID
* src/daemons/rg_forward.c: Give 10 minutes for responses to
forwarded requests.
* src/daemons/rg_state.c: Shorten RG state names. Fix 'Uncertain'
output line.
* src/utils/clustat.c: Fix ccs_member_list() function.
2006-05-23 Lon Hohberger <lhh at redhat.com>
* src/daemons/members.c: Zap pad fields on copy-out
* src/daemons/main.c: Give notice if skipping an event because of
locked services. Call the self-watchdog init function
* src/daemons/watchdog.c: Add Stanko Kupcevic's self-watchdog from
CVS head (fixes #193247)
* src/daemons/groups.c: Add debug messages. Actually count
resgroups during node transition handling
* src/daemons/rg_state.c: allow failover of stopping services if
the owner died (#193255)
* src/utils/clustat.c: fix typo, misc. usability problems (#192999)
2006-05-16 Lon Hohberger <lhh at redhat.com>
* src/resources/nfsclient.sh: Fix 189218 - nfsclient not matching
wildcards correctly when checking status. Allow disabling of
recovery for services where the nfs clients are ordered (this will
cause a full service restart, but works)
* src/resources/clusterfs.sh, fs.sh, svclib_nfslock, service.sh:
Implement rudimentary atomic bomb-style NFS lock reclaim handling
Needs compatible and correctly configured version of nfs-utils
installed and running on the system. For clusterfs.sh, ensure
that we flush buffers during service tear-down - regardless of
whether or not we unmount the file system.
* src/utils/clunfslock.sh: HA-callout program (/usr/sbin/clunfslock)
for use with the rpc.statd -H parameter. Copies the client to all
cluster-managed mounted file systems so that it will get lock
reclaim notification on failover.
2006-05-09 Lon Hohberger <lhh at redhat.com>
* include/list.h: Prevent dereferencing curr if it's null for some
reason
* include/resgroup.h: Clean up alignment, add rgmanager lock/unlock
message types
* src/daemons/Makefile: Add nodeevent.o to the build for rgmanager
* src/clulib/msgsimple.c: Misc code path cleanups
* src/clulib/vft.c: Add local reads for fast clustat operation.
* src/daemons/groups.c: Count all resource groups for all nodes
in one pass, rather than one node per pass. Split queueing of
status checks off so we never block the main thread. Mark services
which have autostart=0 in the config as "disabled" to help remove
confusion between "disabled", "stopped", and the no-longer-needed
"stopped but behave like disabled" states. bz #182454 /
#190234 / #190408
* src/daemons/fo_domain.c: Add patch from Josef Whiter to
implement no-failback option for a given FO domain - bz #189841
* src/daemons/main.c: Queue node events for another thread to
handle, so we never block the main thread. Also, implement
cluster-wide service lock/unlock feature from clumanager 1.2.x
- bz #175010
* src/daemons/nodeevent.c: Split out node event queueing / handling
in to a separate thread so the main thread does not block
* src/daemons/rg_state.c: Return error codes if resource groups
are locked.
* src/daemons/rg_thread.c: Fix assertion failure causing segfault
in extremely rare cases. Quash the rg queue during shutdown.
- bz #181539
* src/daemons/rg_state.c: Add fast local service state query to
reduce unnecessary lock contention
* src/daemons/groups.c: Handle request for expedited information
from clustat.
* src/daemons/main.c: Pass arg1 to send_rg_states() to enable fast
clustat operation.
* src/resources/fs.sh: Implement user/group quota support if
enabled in the file system options
* src/utils/clustat.c: Misc. error handling. Add single service /
member output and add -Q to the help information. #185952.
Added -f flag.
* src/utils/clusvcadm.c: Implement client-side of #175010
* src/utils/clustat.c: show transition time in clustat -x
- bz #191398
* src/resources/fs.sh: enable user/group quotas if enabled in the
options attribute - bz #191182
* init.d/rgmanager: fix typo - bz #191205
-------------
2005-03-21 Lon Hohberger <lhh at redhat.com>
* init.d/rgmanager, Makefile: Fix up init script and add Makefile
so that the init script is properly installed #142754
* src/daemons/*: Fixes for #150344, #151187: Relocate to same node
returns failure, hang during shutdown if user relocate is in-flight.
Fix service state getting stuck in "recoverable" on fail-to-start
scenarios where other nodes failed (or no other node was available)
Rename "resourcegroup" to "service" to be consistent with UI
* src/resources/fs.sh, clusterfs.sh: Fix #151077: Force unmount broken
* src/resources/netfs.sh: Fix #151091: netfs status broken
* src/resources/resourcegroup.sh, service.sh: Remove resourcegroup,
rename to service.sh
2005-03-14 Lon Hohberger <lhh at redhat.com>
* src/resources/clusterfs.sh, fs.sh: Make clusterfs actually work.
Clean up fs.sh + clusterfs.sh "status" when mount reports none/
devpets/usbdev/etc.
* src/daemons/test.c: Add a 'rules' test function for printing
resource rules to stdout.
* src/daemons/reslist.c: Fix 151095
2005-03-07 Lon Hohberger <lhh at redhat.com>
* include/resgroup.h: Add STOP_USER so we can handle user
STOP (instead of just DISABLE) requests. #150333
* src/resources/fs.sh: umount should umount mount points, not
devices. Handle symlinks to file system block devices. #150481
* src/clulib/rg_strings.c: Add user stop string.
* src/clulib/gettid.c: errno fix from trunk
* src/clulib/vft.c: Connect timeout extension for VF
* src/daemons/main.c: Separate connect + login. GuLM doesn't know
about SGs.
* src/daemons/rg_state.c: Change stop handling. Add generic recover
function.
* src/daemons/rg_thread.c: Add support for RESTART, USER_STOP.
#150330, #150333
* src/utils/clusvcadm.c: Use USER_STOP to signal a user-called stop.
#150333
2005-03-02 Lon Hohberger <lhh at redhat.com>
* include/clulog.h: Change default log level to INFO
* include/resgroup.h: Add proto for "best_target_node"
* src/clulib/clulog.c: Change log facility to LOG_DAEMON to match
other cluster daemons (e.g. lock_gulmd)
* src/daemons/groups.c: Add best_target_node, count_resource_groups.
Implement missing autostart-disable feature and requested exclusive
resource group feature. Store configuration view number so we can
tell when the configuration changes.
* src/daemons/main.c: Print node state transition messages before
calling node_event(). Use do_status_checks() so we don't try to
check services we're not running. Bump periodic status event
queueing to 10 seconds instead of 5. Poll ccsd for config updates
since we have no other way to find them. Fix bug preventing
status checks when clustat -i 1 is running.
* src/daemons/rg_state.c: Fix handle_relocate_req so that it uses
best_target_node() correctly. Leave services which failed on all
current nodes as 'stopped', so the next node transition will cause
us to try to restart it automagically. Consider recovery policy
when taking recovery action.
* src/daemons/rg_thread.c: Use recovery routine instead of start.
* src/daemons/restree.c: Fix tree delta updates.
* src/resources/resourcegroup.sh: Add 'exclusive' parameter. Change
'autostart' to a boolean instead of string. Add recovery policy
parameter.
* src/utils/clusvcadm.c: Make "relocate to node X" work.
2005-02-28 Lon Hohberger <lhh at redhat.com>
* errors.txt: Remove random whitespace at the bottom.
* include/resgroup.h: Add do_status_checks proto
* include/rg_queue.h: Remove __ definitions so as not to conflict
with glibc internals.
* include/vf.h: Increase VF_COORD_TIMEOUT to something reasonable.
* src/daemons/groups.c: Add do_status_checks(). We were previously
queueing status checks for RGs that we didn't own. Not useful.
* src/daemons/main.c: Fix for #149410.
* src/daemons/rg_state.c: Fix various failover service problems.
* src/resources/script.sh: Remove "recover" from generic script
wrapper.
=======================================================================
2004-09-23 Lon Hohberger <lhh at redhat.com>
* include/reslist.h: Add needstart/needstop/common flags for
reconfiguration. Added RS_CONDSTART/CONDSTOP to perform "stop
if needed/start if needed" operations after a resource [group]
reconfiguration. Cleaned up structures. Added NO_CCS define for
testing.
* src/daemons/fo_domain.c: Added NO_CCS defines for testing.
* src/daemons/main.c: Added reconfigure() stub function. Added
testing support.
* src/daemons/reslist.c: Added comparsion + primary-attr functions
for resources. Add printout of needstart/stop in resource dump.
* src/daemons/restree.c: Added resource list comparsion +
resource tree comparison functions. Added condstart/stop to ops
list. Added CONDSTART/CONDSTOP handling in res_op.
* src/daemons/rg_locks.c: Added NO_CCS support for testing.
2004-09-13 Lon Hohberger <lhh at redhat.com>
* include/resgroup.h: Add a default check interval.
* include/reslist.h: Add a recover operation, and put operations
and checks together in each instantiated resource structure.
* src/daemons/groups.c: Don't use the old rg_status() func --
its internals have changed.
* src/daemons/reslist.c: Duplicate the action structure of a
parent resource type into an instantiated resource.
* src/daemons/resrules.c: Find the actions with the correct
path.
* src/daemons/restree.c: Add depth parameter to res_exec. Add
do_status - find the highest check/status level to perform given
the elapsed time since another status operation was performed.
Add a reference count each time a resource is started on a node.
* src/daemons/rg_thread.c: Implement periodic status checks.
Currently (in contrast to clumanager 1.2), these status checks
are automatic and not configurable.
* src/resources/*: Misc updates re: check intervals, new
parameters, etc.
2004-09-07 Lon Hohberger <lhh at redhat.com>
* src/resources/group.sh: Add 'autostart' parameter to group
entity
* src/daemons/*: Add support for OCF 'action' specifications.
2004-08-30 Lon Hohberger <lhh at redhat.com>
* src/resources/*: Add status/monitor actions to metadata
* include/list.h: Update to fix compiler warnings. This is not
complete; it's better to add a 'field' to structures requiring
list specs.
* src/clulib/vft.c: Remove unnecessary pthread locks.
* src/daemons/*: Misc. code cleanups.
2004-08-12 Lon Hohberger <lhh at redhat.com>
* global: prepare for RPM build

File Metadata

Mime Type
text/plain
Expires
Sun, Jul 20, 9:18 PM (17 h, 27 s)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
2081519
Default Alt Text
ChangeLog (11 KB)

Event Timeline