> Postgre 7.0.3, on RedHat Linux 6.2 stock 2.2.16 kernel. Nothing special I
> can think of, this server has been up and in use for the last 128 days
> with
> no problem. Last night while cron was performing the nightly vacuuming of
> all databases on one of our servers, I got this from cron.
>
> Vacuuming cms
> FATAL 1: _bt_restscan: my bits moved right off the end of the world!
> Recreate index history_id_key.
> pqReadData() -- backend closed the channel unexpectedly.[snip ]
> I have since stopped the database server and all my users are dead in the
> water at the moment. I took postgres down to single user mode and I'm
> doing
> a vacuum and was considering doing an iccpclean. Any other suggestions?
> dump & restore? Any Idea what happened?
>
The vacuum I tried in single user mode failed (froze on a table, for over 20
minutes) so I killed it. I cleaned up shared memory (there were some things
left over). I started in single user mode and reindex database cms force,
also reindexed a few tables, then tried the vacuum again with the same
result. Do I need to dump resore? We should have a valid backup from the
night before.
