Re: index and table corruption

Hi Jerry,

Thanks for the suggestion

Yes, until about a month ago, we weren't wrapping our snapshots with
pg_start_backup and pg_stop_backup. Same reason as you mentioned, the
database would start up and "trivial checks" would be okay, and so we
figured "why write a script?".

However we did change that a month or so ago ago, and have had the problem
after that. Every snapshot we have tried to actually recover from has been
wrapped in a pg_start_backup and pg_stop_backup, so we are leaning more
towards server/disk corruption at this time.

We also synced our snapshot to an alternate SAN, and ran a script to
update every row of every table, and do a full vacuum and reindex of every
table, and there were no error messages about bad blocks.

Thanks,
Karthik