Re: Minimally avoiding Transaction Wraparound in VLDBs

Christopher Kings-Lynne <chriskl(at)familyhealth(dot)com(dot)au> writes:
>> I really really do not like proposals to introduce still another kind
>> of VACUUM. We have too many already; any casual glance through the
>> archives will show that most PG users don't have a grip on when to use
>> VACUUM FULL vs VACUUM. Throwing in some more types will make that
>> problem exponentially worse.

> Yes, but if they're all under the control of autovacuum, then users
> don't have to worry...

Well, if the proposal comes packaged with an algorithm by which
autovacuum will use it, that's a different story. What's sticking in
my craw about this proposal is really that it's assuming detailed manual
management of vacuuming, which is exactly the thing we've been sweating
to get rid of.

BTW ... the original Berkeley papers on Postgres make frequent reference
to a "vacuum daemon", which seems to be essentially what we're trying to
build with autovacuum. Does anyone know if the Berkeley implementation
ever actually had auto vacuuming, or was that all handwaving? If it did
exist, why was it removed?

regards, tom lane