Re: Reaping Temp tables to avoid XID wraparound

On Wed, Feb 13, 2019 at 6:05 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:

Magnus Hagander <magnus@hagander.net> writes:
> And while at it, what would in this particular case have been even more
> useful to the OP would be to actually identify that there is a temp table
> *and which xid it's blocking at*. For regular transactions we can look at
> backend_xid, but IIRC that doesn't work for temp tables (unless they are
> inside a transaction). Maybe we can find a way to expose that type of
> relevant information at a similar level while poking around that code?

Maybe I'm confused, but doesn't the table's pg_class row tell you what
you need to know?  You can't look inside another session's temp table,
but you don't need to.

I believe it does, yes.

But that doesn't make for a way to conveniently go "what is it that's causing waparound problems", since due to pg_class being per database, you have to loop over all your databases to find that query. Having that information available in a way that's easy for monitoring to get at (much as the backend_xid field in pg_stat_activity can help you wrt general snapshots) would be useful.

--

 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/