Subject | really strange problem with IB server 5.6 |
---|---|
Author | Stuart Hunt |
Post date | 2002-10-22T16:00:52Z |
Hi,
We've got a really peculiar and damaging problem at one customer's site only
(out of about 400) with BLOb fields.
The user started reporting "data error" errors being returned occasionally
when trying to view certain blobs for about a week. In itself I've never
heard of this before.
All of a sudden they started getting "unassigned code" and "wrong page type.
page 6071 is of wrong type (expected 8, found 5)" with 4 BLObs (out of
approx. 130). All the others are fine, and the really strange is 3 out of
these 4 haven't been modified in weeks whereas several of the OK ones have.
We have attempted to reproduce this problem using their data from before the
problem and cannot. However, their current data produces this problem all
the time and will not backup. When you try you get a gds_get_segment$
failure and the backup stops. The only postings I've come across state that
the DB is corrupt and gfix -mend must be used to remove the corrupt records
or go back to a previous backup.
The problem is that the customer is demanding to know why this is happening
and wants assurances it won't happen again.
does anyone out there know what can cause these BLOb corruptions and what to
do to stop them?
The pecularities of this customer's site is that the are viewing over a VPN
and using an IB-server side UDF to encrypt the BLObs because of this.
However, we have used this UDF ourselves in a multi-user environment without
any problems at all.
Any help much appreciated,
Stuart Hunt,
Technical Leader,
IdeaGen Software PLC
We've got a really peculiar and damaging problem at one customer's site only
(out of about 400) with BLOb fields.
The user started reporting "data error" errors being returned occasionally
when trying to view certain blobs for about a week. In itself I've never
heard of this before.
All of a sudden they started getting "unassigned code" and "wrong page type.
page 6071 is of wrong type (expected 8, found 5)" with 4 BLObs (out of
approx. 130). All the others are fine, and the really strange is 3 out of
these 4 haven't been modified in weeks whereas several of the OK ones have.
We have attempted to reproduce this problem using their data from before the
problem and cannot. However, their current data produces this problem all
the time and will not backup. When you try you get a gds_get_segment$
failure and the backup stops. The only postings I've come across state that
the DB is corrupt and gfix -mend must be used to remove the corrupt records
or go back to a previous backup.
The problem is that the customer is demanding to know why this is happening
and wants assurances it won't happen again.
does anyone out there know what can cause these BLOb corruptions and what to
do to stop them?
The pecularities of this customer's site is that the are viewing over a VPN
and using an IB-server side UDF to encrypt the BLObs because of this.
However, we have used this UDF ourselves in a multi-user environment without
any problems at all.
Any help much appreciated,
Stuart Hunt,
Technical Leader,
IdeaGen Software PLC