Re: Missing Chunk Error when doing a VACUUM FULL operation - DB Corruption?

Arjun Ranade <ranade(at)nodalexchange(dot)com> writes:
> There is nothing in pg_prepared_xacts, however, in pg_stat_activity there
> are two pglogical processes that have a "backend_start" of "2017-06-17"
> when the last time we restarted this server. Both are not waiting and have
> null for "state." This might be expected behavior for pglogical.

I don't know much about pglogical, but if it's holding back the xmin
horizon (as it appears to be doing) that is a really bad pglogical bug.

If you try vacuum verbose (doesn't have to be FULL) on some table that
gets lots of update traffic, does it report a large number of dead-but-
not-removable rows?

> I did try TRUNCATE before but even as the postgres user, but for some
> reason it didn't allow me to truncate because it was a system table.

Mmm. You could get around that, but possibly not without restarting
in single-user mode, which is probably more interruption in service
than you want. It might be easier to force-quit the pglogical sessions.

regards, tom lane