Thanks Tom. OK, I ran: postgres -O -P -D /cluster/location
reindex table pg_class; backend> reindex table pg_class; And then nothing returns. Nothing stating whether it was successful or a failure. And then if I do: bash-3.2$ postgres -O -P -D /shared/webapp/database webapp I still get: FATAL: cache lookup failed for index 2662 On Wed, Jun 30, 2010 at 12:14 PM, Tom Lane <[email protected]> wrote: > "Kevin Grittner" <[email protected]> writes: > > OK, I think the appropriate next step would be to try to run the > > PostgreSQL cluster in single-user mode: > > > http://www.postgresql.org/docs/8.1/interactive/app-postgres.html > > > Try to REINDEX pg_class_oid_index in that mode. If that fails, it > > might possibly help to run these statements and try the REINDEX > > command again: > > > set enable_indexscan = off; > > set enable_bitmapscan = off; > > Those won't help. What you *will* need, in order to even start the > single-user backend, is to tell it to disregard system indexes > (-P command line option). > > I wouldn't be too surprised if the corruption extends a lot further than > the one index :-( but maybe you will be able to extract something after > reindexing. > > regards, tom lane >
