Implement cleanup mechanism for DB queues when mount process dies unexpectedly
Currently the DRIVE_STATE
table in the catalogue holds the sessionId
which is mapped to the mountId
of the mount which it operated on. We can use this to check all DB jobs which have had no progress since some configurable timeout if they have an active session ID in the drive state table. If not, we can simply requeue the job back to the insert queue table.
Edited by Jaroslav Guenther