3.7.0 HP StorageWorks HP Scalable NAS File Serving Software release notes for Linux (AG513-96006, October 2009)

DescriptionDefect
Administrative filesystem should be excluded from scans
The 3.7 release includes an administrative filesystem (adminfs) that is used to store cluster
operational data. HP recommends that the parts of the administrative filesystem that are used
for replication or performance dashboard storage not be marked for real time scanning by
any anti-virus or malware software you may be running on the cluster. Scanning the
administrative filesystem replication directories can materially impact the speed of replication
in the system. No user data is stored on the administrative filesystem in the replication or
performance dashboard directories.
The default directory for performance dashboard historical data is ${ADMINFS_DIR}/
perfmon/ganglia/rrds/.
The default replication directory is ${ADMINFS_DIR}/replication/.
HP recommends that these directories be excluded from any AV scan. If possible under the
security rules for the customer site, the entire administrative filesystem should be excluded
from scan; this can improve replication performance.
27464
rplstatus reports status incorrectly when servers are down
When a server is fenced or shuts down unexpectedly, the rplstatus utility incorrectly
reports that the server status is “running.”
27467
Replication alerts cannot be cleared from the Management Console or command line
Alerts issued by replication programs cannot be cleared either on the Management Console
or with replication commands.
Workaround. Restart replication on the node where the alert was issued. The alert will then
be cleared.
27704
rplstatus reports that it cannot read the mxds key
When you add or remove servers in the cluster, the rplstatus command may report the
following:
Error: Can't read mxds key /Replication/config/1/currentsentinel
This error does not affect replication operations; replication will continue to work correctly.
To remove the message, reimport the configuration:
rplconfig -e <current_config_file
rplconfig -i <current_config_file>
The rplstatus command will now report the correct sentinel node.
28059
Files and directories might not be deleted on the destination cluster when replication is stopped
In certain circumstances, files and directories that are deleted on the source cluster might not
be deleted from the destination cluster. If replication is stopped on the source cluster before
deletes are communicated to the destination cluster, those delete actions will not be replicated
upon restarting replication. This is unlike file and directory creation and modification, which
will be synchronized with the destination upon restarting replication.
28439
rplcontrol -s restart command will not restart replication after a node failure
When a node fails and is then rebooted, replication may be in the stopped state on the node.
Normally the rplcontrol -s restart command is used to change the replication state from
stopped to running; however, the command may not work after a node failure.
Workaround. Run the following commands to restart replication on the node:
rplcontrol -s exit -h <host ip>
rplcontrol -s enable -h <host ip>
/etc/init.d/pmxs start
28441
12