On Thu, May 7, 2015 at 12:23 AM, Thomas Munro
<thomas.munro@enterprisedb.com> wrote:
> This patch will change it anyway, but I noticed that oldestOffset's
> computation to find the start of the segment seems wrong in master, I
> think it should be like this, no?
>
> @@ -2495,7 +2495,7 @@ DetermineSafeOldestOffset(MultiXactId oldestMXact)
> */
> oldestOffset = find_multixact_start(oldestMXact);
> /* move back to start of the corresponding segment */
> - oldestOffset -= oldestOffset / MULTIXACT_MEMBERS_PER_PAGE *
> SLRU_PAGES_PER_SEGMENT;
> + oldestOffset -= oldestOffset % (MULTIXACT_MEMBERS_PER_PAGE *
> SLRU_PAGES_PER_SEGMENT);
>
> LWLockAcquire(MultiXactGenLock, LW_EXCLUSIVE);
> /* always leave one segment before the wraparound point */
This should be committed and back-patched separately, I think. I'll
go do that now.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company