I'm running FB 1.5 classic on a Win2k3 server, with an 18GB database.

We backup and restore to a temp folder every night. This past Friday
the backup failed, with no information as to why - the gbak log just
stopped in mid-sentence, and the FB process associated with it blew
up.

After running gfix, I found that a number of indices were corrupt. I
rebuilt those, and tried to backup again. The backup still failed. I
finally tried a backup with the -g -ig -nt switches, and was
succesful. After the successful restore, I replaced the production
database.

Last night was the first attempt at the normal backup/restore routine,
and it failed in the same way as this past Friday. I did not have the
luxury of playing around today, so I just tried a gbak with the -ig
-nt switches. Surprisingly, the backup was successful.

My question is this - what does it tell me that a backup with the -ig
and -nt is successful, but without them is not successful? I cannot
tell if it was the -ig or the -nt switch that allowed succesful
completion (due to time restrictions). Does anyone have a guess which
might be the culprit?

tia,

Bob M..