[2456] in SIPB-AFS-requests
that backup isn't happening
daemon@ATHENA.MIT.EDU (Sam Hartman)
Wed Sep 18 02:37:18 1996
Date: Wed, 18 Sep 1996 02:37:03 -0400
From: Sam Hartman <hartmans@MIT.EDU>
To: sipb-afsreq@MIT.EDU
So, the AFS backup didn't want to happen today. Basically,
the current problem is that the DLT drive doesn't want to work
reliably enough to do a backup.
At about 18:47 the backup I started at noon failed because of
an unexpected SCSI bus reset. I talked to tlyu, and he suggested
shutting down btc and power cycling all devices on the SCSI chain. I
did this.
The backup promptly failed with a tape I/O error so I decided
to try writing out a few hundred megs of data to a type and reading it
back. I dd'd one of btc's disk partions to the tape and read it into
/dev/null without a problem.
Things still failed, so I shut down btc and checked all the
SCSI cables to see if they were connected properly. I rebooted btc
and tried a backup.
I could not get sipb-all.full3 too load. The tape drive
randomly resets (power cycles or something) during the middle of
loading the tape.
I tried loading a new tape which also failed, but I'm not
convinced that wasn't caused by the tape drive being confused. I
power cycled again, got it to load sipb-all.full3 but the backup
failed.
I'm now going to try power cycling and loading a new tape.
However, I think we should consider getting the DLT looked at, and
moving the tapeconfig entry for the Exabyte to btc in case we have to
fail back to Exabyte backups for a week or so.
--Sam
