On Mon, Sep 12, 2011 at 9:41 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Dylan Adams <dylan.adams.work@gmail.com> writes:
>> [ persistent occurrences of index corruption ]
>
>> My primary question: is this normal?
>
> No. It does sound like you're managing to tickle some bug or other.
> Can you extract a test case of any kind? We could fix it if we could
> see it happening, but there's not enough information here for that.
I haven't been able to come up with a self contained test case.
There have been a few instances where a particular series of batch
processes which, when run repeatedly on a particular data set, will
reproduce the problem consistently. But it's not possible to release
the required code and data.
dylan