On Wed, Dec 29, 2010 at 1:53 PM, bricklen <bricklen@gmail.com> wrote:
> On Wed, Dec 29, 2010 at 12:11 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>>
>> The difference in ctid, and the values of xmin and relfrozenxid,
>> seems to confirm my suspicion that this wasn't just random cosmic rays.
>> You did something on the source DB that rewrote the table with a new
>> relfilenode (possibly CLUSTER or some form of ALTER TABLE; plain VACUUM
>> or ANALYZE wouldn't do it). And for some reason the standby hasn't
>> picked up that change in the pg_class row. I suspect the explanation
>> is that your technique for setting up the standby is flawed. You can't
>> just rsync and have a valid snapshot of the DB --- you need to be sure
>> that enough WAL gets replayed to fix any inconsistencies arising from
>> the time-extended nature of the rsync operation. But you didn't say
>> exactly how you did that.
>>
>
> Definitely no CLUSTER commands were issued, and there should have been
> no ALTER commands issued (yesterday was a holiday, no one was here).
> Would a TRUNCATE have the same effect though? I grep'd through our
> application, and it appears that at least 3 tables get truncated, one
> of them several times per hour. The often-truncated table wasn't one
> of the bad ones, but the others are the ones I've already identified
> as non-existent.
Update: Set up the warm standby again and encountered the same issue,
with two of the three previously-identified tables -- the ones that
can get truncated throughout the day. We're going to try again
overnight when those tables are not truncated and see if that gives us
a correctly-working standby.
From what I could find from posts to these lists, TRUNCATE commands do
reset the relfilenode, and that could account for the issue we are
experiencing. What I find odd is that we have one other table that is
truncated every 15 minutes (aggregate table) but that one was fine in
both attempts at the warm standby.
