On Thu, Jun 4, 2015 at 5:29 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> - Forces aggressive autovacuuming when the control file's
> oldestMultiXid doesn't point to a valid MultiXact and enables member
> wraparound at the next checkpoint following the correction of that
> problem.
Err, enables member wraparound *protection* at the next checkpoint,
not the wraparound itself.
It's worth noting that every startup will now include one of the
following two messages:
LOG: MultiXact member wraparound protections are now enabled
Or:
LOG: MultiXact member wraparound protections are disabled because
oldest checkpointed MultiXact %u does not exist on disk
...where %u is probably 1
If you get the second one, you'll get the first one later after vacuum
has done its thing and a checkpoint has happened.
This is, obviously, some log chatter for people who don't have a
problem and never have, but I think it's worth emitting the first
message at startup even when there's no problem, so that people don't
have to make inferences from the absence of a message. We can tell
people very simply that (1) if they see the first message, everything
is fine; (2) if they see the second message, autovacuum is going to
clean things up and they will eventually see the first message; and
(3) if they see neither message, they haven't upgraded to a fixed
version yet.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company