Issues fixed in VCS 5.0 MP3

Table: VCS 5.0 MP3 fixed issues lists the fixed incidents for VCS 5.0 MP3.

Table: VCS 5.0 MP3 fixed issues

Incident

Fix description

612587

The haclus -wait command does not hang now when cluster name is not specified.

618961

On SUSE nodes, when the fencing driver calls the kernel panic routine, this routine could get stuck in sync_sys() call and cause the panic to hang. This allows both sides of the split-brain to remain alive. This issue is fixed so that the cluster node panics after split-brain.

797703

The output of the vxfenadm command with -d option had unwanted "^M" character attached to the RFSM state information of all nodes. The output of the vxfenadm command now does not display the unwanted "^M" characters.

805121

Partial groups go online erroneously if you kill and restart the VCS engine.

861792

Service groups are now listed in the groups table in the Recover Site wizard.

862507

GAB_F_SEQBUSY is now not set when the sequence request is not sent.

896781

The issue that caused systems to panic intermittently after you disable array side switch port is fixed.

900244

When a service group switches over, NFS clean is not called anymore.

914752

Optimized LLT to print "node inactive" messages less frequently so as to not flood the dmesg buffer.

926849

Fixed an issue that prevented service groups with IPMultiNIC resources from failing back.

989935

The clean and monitor functions of the Application agent can now detect the same process/application run by different users.

1016548

The issue that caused node panic with message "GAB: Port f halting system due to network failure" is fixed.

1032572

The vxfen-startup script is modified to use the xpg4 awk utility (/usr/xpg4/ bin/awk).

1038373

Fixed an issue where LLT was causing panic because it was referencing freed memory.

1045128

Resolved an issue that caused gab to panic during initialization.

1050999

Resolved a HAD issue that caused the ShutdownTimeout value to not work correctly.

1051193

The vxfen unconfigure command now uses the correct key to preempt/abort from now faulted paths.

1053377

/usr/bin:/bin:/usr/sbin:/sbin is added first in the PATH env variable so that 'df', 'uname', and so on are picked up from these paths.

1055379

vxfentsthw works without error when raw device file is not present .

1056559

Fixed an issue where the NFSRestart monitor threw a, "too many open files error."

1057418

The I/O fencing component can now retrieve the serial number of LUNs from a Pillar Data array. I/O fencing can now start without any issues and port b comes up for PillarData Arrays.

1061056

Child process of the agents now do not make calls to the logging functions of the VCSAgLog object.

1067667

When the VCS engine is stopping with evacuation, even if resource faults for a group with OnlineRetryLimit greater than zero, the evacuation completes.

1096394

Fixed an issue where notifier blocked the IPM socket.

1099651

Fixed the issue where the Process agent flooded the log and message files with the V-16-1001-9004 error message.

1102457

The vxfen initialization code has been modified to properly initialize parameters and avoid dereferencing of null pointer.

1113667

Replaced a fork call with a safer one.

1133171

When new resources are added, the service threads in the corresponding agent will get created upto a maximum of NumThreads threads.

1133223

All the thread unsafe calls are replaced with thread safe calls.

1137118

The fencing driver retries startup for some time before deciding on a pre-existing split brain.

1156189

The Sybase resource can now come online even if - s appears in the path of the server name of the resource attribute definition.

1162291

Application agent now inherits user defined LANG parameter.

1174520

The VCS engine now does not assert if a service group with single system in SystemList is unfreezed.

1174911

Group switch/failover logic now completes if parent group gets autodisabled.

1186414

The hastart command and the triggers now run on the locale specified by the LANG variable.

1187580

Issues with the ActionTimeout attribute are fixed.

1189542

Service Group state now does not turn from PARTIAL to ONLINE after killing the VCS engine. The issue is fixed to show the correct state of the service group.

1193697

The default heartbeating mechanism in LLT is now point-to-point and not broadcast heartbeating.

1205904

Mount agent's problem to monitor two Mount resources with same MountPoint but different BlockDevice is fixed.

1214464

The VCS agent for Sybase now uses getent passwd to check the user details defined in the Owner attribute.

1217482

Fixed the cluster_connector.sh script to use the correct signal to stop the Java process when the CMC group goes offline.

1218881

The Apache version was incorrectly parsed for IBMIHS.

1228409

With UmountVolumes attribute of DiskGroup agent set to 1, the resource can failover with open volumes outside the VCS control.

1228356

The VCS agent for DB2 now uses getent passwd to check the user details defined in the Owner attribute.

1230862

The nfs_postoffline trigger can now read the value of the NFS agent's Nservers attribute.

1241260

OnlineRetryCount, OnlineWaitCount, and OfflineWaitCount values are reset when a resource is flushed.

1247347

The Mount agent error message now indicates that there are trailing spaces in the BlockDevice or MountPoint attribute value. The user must remove these extra spaces for the resource to be correctly configured.

1256313

While unfreezing a group, check if there exists concurrency violation. Fire a concurrency violation trigger to remove the concurrency.

1259756

When NFSv4Support is set to 0, the nfsd and mountd daemons start with "--no-nfs-version 4" option.

1268550

The DiskGroup agent is enhanced to support large storage configurations.

1280144

Subroutines are added to set and reset locale in the ag_i18n_inc module.

1285439

The clean script of the VCS agent for Sybase now supports removal of the IPC resources that the Sybase processes allocate.

1296465

If the ClusterService group is online on the same node on which a global service group is also online and if the node goes down, then the service group is failed over to a remote site.

1298642

The Mount agent umounts all the bindfs filesystems associated with the mount point before taking the resource offline.

1379532

The online script of the VCS agent for Oracle is fixed to use the startup force command when the value of the StartUpOpt attribute is set to STARTUP_FORCE.