During a vacuum, I ran into this error:
vacuumdb: vacuuming of database "friend" failed: ERROR: invalid page
header in block 41661 of relation "friend_pkey"
I've read the posts on this newsgroup and it's clear that I have to
REINDEX to fix this. The bigger question is -- why did this happen in
the first place? I'm using open_sync as my WAL_SYNC_METHOD. Is it
better to use fsync?
Running Suse 9.0 on a Quad Opteron, with direct attached storage. The
RAID card is set to write through all of the writes, so I don't think
this is a case where the RAID card is lying about a commit.
-Akash