Open System Services Management and Operations Guide (G06.30+, H06.08+, J06.03+)

If the OSS Monitor finds that a fileset was left in the STARTED state but its OSS name server process
is not running, one of the following might have occurred:
• A serious problem has occurred.
• A fileset was not properly stopped before the OSS Monitor last stopped.
• An OSS name server failed while the OSS Monitor was not running.
That fileset might need repair.
If the OSS Monitor finds that a fileset was left in the UNKNOWN state, its OSS name server might
have failed while the OSS Monitor was not running. Again, that fileset might need repair.
The OSS Monitor then attempts to restart each fileset suspected of needing repair, in the order of
the fileset mount points within the OSS file-system directory structure, beginning with the root fileset.
If the OSS name server for a fileset reports that the fileset cannot be restarted, the OSS Monitor
runs the Guardian FSCK utility, then tries again to restart the fileset.
The OSS Monitor restart sequence does not wait for the FSCK repair operation on a fileset to finish;
the restart sequences continues with other filesets. The OSS Monitor does wait indefinitely for a
repair operation to finish before attempting to restart the fileset. If the attempt to restart the fileset
fails, the OSS Monitor marks the fileset state as UNKNOWN.
If the OSS Monitor is restarted while an OSS name server is running, the new instance of the OSS
Monitor continues to monitor the OSS name server and can recover from any future failures of the
OSS name server.
Automatic Restart of Filesets by the Automatic Startup Service
A fileset can be configured so that the OSS Monitor automatically starts that fileset after a system
load, regardless of whether the fileset was in the STARTED state. Restarted filesets are automatically
repaired if necessary.
The automatic startup service can also restart the fileset a maximum number of times during a
10-minute period. See the “ADD FILESET Command” (page 256) or the “ALTER FILESET Command”
(page 269) for more information about this service.
Automatic Restart of Filesets After OSS Name Server Failure
If an OSS name server fails, the OSS Monitor initiates a recovery procedure similar to that performed
during OSS Monitor startup. All filesets that were left in the STARTED state and were managed by
that OSS name server (as the OSS name server for either the fileset or the mount-point) are repaired
and restarted. The OSS Monitor also restarts that OSS name server.
Automatic Restart of OSS Name Servers After Processor Failure
If failure of a processor causes failure of an OSS name server process running without a backup
or causes termination of a fault-tolerant OSS name server process pair, the OSS Monitor initiates
a recovery procedure similar to that performed upon OSS Monitor startup. No recovery is attempted
if the processor failure only affects one process of the OSS name server process pair.
During the recovery, all filesets that were left in the STARTED state and were managed by the failed
OSS name server (as the OSS name server for either the fileset or the mount-point) are repaired
and restarted. The OSS Monitor also restarts that OSS name server.
Potential Problems During Automatic Restart of Filesets
The OSS Monitor might be unable to successfully restart all filesets that were left in the STARTED
state when their OSS name servers failed. If a failure occurs, you can attempt to manually recover
from that failure.
NOTE: When the OSS Monitor attempts automatic restart of filesets, it does not retry if certain
failures occur.
148 Managing Filesets