On Jul 26, 2018, at 12:14 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Jeff Frost <jeff@pgexperts.com> writes:
>> On Jul 26, 2018, at 9:50 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>>> My first thought would be something to do with a restrictive search_path.
>
>> Good thought. As far as I can tell we never change it from the default and everything in that DB is in the public
schema.
>
> Right, but what I'm half-theorizing is that the patches for CVE-2018-1058
> affected you somehow. Maybe something "transiently" installed pg_catalog
> (only) as search_path and forgot to remove it again. It would be easy
> to explain it that way if the errors had been coming out of autovacuum
> worker processes, for instance, though your log didn't match that exactly.
>
> If you see it again it would be a good idea to look very closely at what
> happened right before the errors started.
Oh, I see what you're saying. Don't the autovacuum processes connect to their own backends which then exit after
they'recomplete though?
I have the logs around, so if you know what I should look for, I'd be happy to dig.