Re: Re: [GENERAL] 9.4.1 -> 9.4.2 problem: could not access status of transaction 1

Hi,

On 2015-06-04 12:57:42 -0400, Robert Haas wrote:
> + /*
> + * Do we need an emergency autovacuum? If we're not sure, assume yes.
> + */
> + return !oldestOffsetKnown ||
> + (nextOffset - oldestOffset > MULTIXACT_MEMBER_SAFE_THRESHOLD);

I think without teaching autovac about those rules, this might just lead
to lots of autovac processes starting without knowing they should do
something? They know about autovacuum_multixact_freeze_age, but they
know neither about !oldestOffsetKnown nor, afaics, about pending offset
wraparounds?

> -static MultiXactOffset
> -find_multixact_start(MultiXactId multi)
> +static bool
> +find_multixact_start(MultiXactId multi, MultiXactOffset *result)
> {
> MultiXactOffset offset;
> int pageno;
> @@ -2630,6 +2741,9 @@ find_multixact_start(MultiXactId multi)
> pageno = MultiXactIdToOffsetPage(multi);
> entryno = MultiXactIdToOffsetEntry(multi);
>
> + if (!SimpleLruDoesPhysicalPageExist(MultiXactOffsetCtl, pageno))
> + return false;
> +
> /* lock is acquired by SimpleLruReadPage_ReadOnly */
> slotno = SimpleLruReadPage_ReadOnly(MultiXactOffsetCtl, pageno, multi);
> offptr = (MultiXactOffset *) MultiXactOffsetCtl->shared->page_buffer[slotno];
> @@ -2642,25 +2756,31 @@ find_multixact_start(MultiXactId multi)
>

I think it'd be a good idea to also return false in case of a
InvalidMultiXactId - that'll be returned if the page has been zeroed.

Andres