LMT Database Recovery Process

Lustre monitoring tool (LMT) database recovery process after metadata server (MDS) crash on direct-attached Lustre (DAL).

The Lustre monitoring tool (LMT) database can be corrupted when the management server (MGS)/primary metadata server (MDS) crashes in a direct-attach Lustre (DAL) file system. The corruption can be repaired by running mysqlcheck on the MGS/primary MDS.

Run mysqlcheck just after the primary MDS is rebooted. LMT will work as soon as the primary MDS is rebooted so long as the database is usable. If mysqlcheck is run after reboot, performance numbers are generated from LMT even when using the secondary MDS.

nid00325# mysqlcheck -r -A -p
Enter password:
filesystem_dal.EVENT_DATA                          OK
filesystem_dal.EVENT_INFO                          OK
filesystem_dal.FILESYSTEM_AGGREGATE_DAY            OK
filesystem_dal.FILESYSTEM_AGGREGATE_HOUR           OK