Hi,
On 2018-06-28 10:43:16 -0600, Scott Ribe wrote:
> > On Jun 28, 2018, at 10:36 AM, Andres Freund <andres@anarazel.de> wrote:
> >
> > How are you making the inference that that's the problem? It could very
> > well be a postgres bug or somebody having used pg_resetxlog or.. ?
>
> There could always be a postgres bug, however misconfigured cache
> layers are a very common cause of this kind of symptom, and postgres
> bugs very rare.
Sure, but it's also easy to not find postgres bugs if you don't
investigate. If the error comes from parsing WAL - which is likely but
not guaranteed from the log output - then it's unlikely the error is
caused by on-disk corruption. The WAL is protected by a CRC checksum,
and if it fails, you don't get to the memory allocation bit.
> I don't believe pg_resetxlog would cause this particular
> problem. Failure to recovery yes, but not illegal memory operation.
It can. Not as a result of reading WAL itself, but as a result of the
resulting corruption. That's why I asked for debugging output.
- Andres