"Mark Steben" <msteben@xxxxxxxxxxxxxxx> writes:
> The number may change pointing to another index or stay the same.
> The error persists if I repeat 3 or 4 times. Usually after the
> 4th or 5th time the cluster is successful.
Hm. I'd guess that the installation is old enough that the OID counter
has wrapped around, and is currently in a region that is pretty densely
populated with existing tables. (You could check that out by looking
into the database directory.) 7.4.x doesn't have the logic that was
added later to skip over filenames that are already in use; instead
you just get an error if there's a collision.
I forget whether 7.4's pg_resetxlog has an option to set the OID
counter, but if it does then a temporary fix is to move the OID counter
to some less densely populated range.
> What is the risk of performing an upgrade to 7.4.19?
Low, but it won't make this particular problem go away.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?
http://archives.postgresql.org
