We have one full tape backup job that runs every Wednesday night. This job backups two repositories. One of type 'Hyper-V Backup' and one of type 'Windows (agent)'. This job uses a tape from a media pool named "Wed".
We also have one job that runs every morning EXCEPT for Wednesday. This job backups the same two repositories. One of type 'Hyper-V Backup' and one of type 'Windows (agent)'. This job uses a tape from a media pool named "Thur". This job runs a full backup on Thursday and then incremental jobs Fri, Sat, Sun, Mon, and Tues.
We use a standalone tape drive.
On Wednesday night the tape backup job started and retired the tape (faulty tape I presume). The job went into a waiting (Add tape to media pool Wed) state. When we ejected the tape to insert a new tape the job canceled itself stating "TapeDrive alert: The operation has failed because the tape cartridge was manually de-mounted while the tape drive was actively writing or reading..” Is this correct behaviour to cancel itself? Is there any way to avoid the job canceling if we need to eject a tape in order to insert a new (or correct) tape for a job?
Because the person who changed the tape wasn't looking at the console after they ejected the tape on Thursday morning, when they inserted a new tape (so the Wed job could continue) it went in to the free media pool but (1)because that Wed night job auto canceled, (2)because the Thurs media pool was set to use free tapes as needed (we've now disabled this), and (3)because the Thursday morning job was already waiting in the queue, the Thursday morning job ran which is not what we wanted. To fix this we let the job finish, moved the tape from the Thur media pool to the Wed media pool and manually ran the Wed night job. Everything backed up as expected (most recent complete chains).
Then we inserted the Thur tape, and manually ran the Thurs morning job. The 'Hyper-v Backup' repository backed up fine (it backed up the most recent chain of .vbk and .vib files which is what we need). However for the 'Windows (agent)' repository part of the job ONLY the most recent .vib files were backed up and not the most recent chain.
In addition to the two questions above we'd also like to learn how we can manually re-run a job and ensure we always get the most recent complete chain. This question applies to both "Hyper-V Backup of Windows agent repository" and "Files to Tape Backup" jobs.
Thank you everyone!