We are running NBU 7.0.1 with Oracle T10000B tape drives in an SL8500 silo. Our environment consists of one master server and two media servers with the tape drives SAN attached to all three servers. We have 7 drives actually mapped to one media server, 6 to the other, and 2 to the master and are not using the shared storage option. Here is our problem:
One one media server, at seemingly random times, 5 of the 7 tape drives will get an unload request (we see it in the bptm report) but will never actually unload the tape which leaves these drives in an ACTIVE control state in our device monitor but with no recorded or extrenal media label showing. When we look on our ACSLS server, the drives continue to show as loaded and will not let us dismount the tapes as it believes that there is still an active job on them. NBU will eventually fail the backup job with a code 52, timed out waiting to load tape, and then another backup job jumps on the drive and the cycle continues. This only happens on the one media server and only on the same 5 tape drives, the other two drives on the server never have an issue.
In order to get the drives working again, we have had to restart the NetBackup services on the media server, though at times we have had to restart the whole environment, and on a couple of occasions actually had to reboot the server. Has anyone else experienced this? And if so, what could be the possible cause? We can't find any rhyme or reason to it. The zoning on the tape drives looks fine and we have seen no errors in the switch, though the media server error log is showing tons of adapter errors. Though if it were the fibre or the HBA we would expect it to be happening on all 7 drives.
