Per the DBAs, there hadn't been any recent crashes before last Thursday.
A "vacuum analyze verbose" discovered the problem early Thursday
morning. After the PANIC, the database never came back up (the
heap_clean_redo: no block / full_page_writes = off problem).
One thing that seems strange to me is that the original crash on
Thursday failed on Panel_pkey, but my "vacuum analyze verbose" on a copy
of the crashed database failed on MaintCode /
pg_statistic_relid_att_index.
I'll send over pg_statistic_relid_att_index and Panel_pkey. Showing
the keys to Panel_pkey is no problem.
Pete
>>> Tom Lane <tgl@sss.pgh.pa.us> 04/12/06 6:03 pm >>>
You did say that this problem showed up shortly after a database
crash,
right?
Could you send me a copy of pg_statistic_relid_att_index, off-list
(the actual disk file, not a pg_filedump)? There's nothing but table
OIDs and attribute numbers in it, so I can't see any reason anyone
would consider it sensitive data. I've got some really crude code
laying about for scanning an index and looking for inconsistencies ...
it's too ugly to give out, but I'd like to see if it can find anything
wrong with that index. (I'll probably ask for a copy of Panel_pkey
for the same purpose, if you get permission to show us its keys.)
regards, tom lane