Обсуждение: 9.0.15 WAL mis-replay status?
I finally got permission to patch a highly used db from 9.0.15 to .17. Is there any way to know if the WAL replay issue has effected the standby? I don't have the reported bloat, is that THE test? Is there any way to measure the current state of this issue listed in the 9.0.16 release notes:
Fix possible mis-replay of WAL records when some segments of a relation aren't full size (Greg Stark, Tom Lane)
The WAL update could be applied to the wrong page, potentially many pages past where it should have been. Aside from corrupting data, this error has been observed to result in significant "bloat" of standby servers compared to their masters, due to updates being applied far beyond where the end-of-file should have been. This failure mode does not appear to be a significant risk during crash recovery, only when initially synchronizing a standby created from a base backup taken from a quickly-changing master.
Вложения
Ray Stell <stellr@vt.edu> writes:
> I finally got permission to patch a highly used db from 9.0.15 to .17. Is there any way to know if the WAL replay
issuehas effected the standby? I don't have the reported bloat, is that THE test? Is there any way to measure the
currentstate of this issue listed in the 9.0.16 release notes:
> Fix possible mis-replay of WAL records when some segments of a relation aren't full size (Greg Stark, Tom Lane)
TBH, I'd suggest re-syncing the standby after you upgrade. Obvious bloat
is one symptom, but only one symptom.
Note however that the issue could only arise on tables that exceed 1GB, so
depending on your usage, maybe you don't have a problem.
regards, tom lane
On Apr 28, 2014, at 11:07 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > Ray Stell <stellr@vt.edu> writes: >> >> Fix possible mis-replay of WAL records when some segments of a relation aren't full size (Greg Stark, Tom Lane) > .. > Note however that the issue could only arise on tables that exceed 1GB I assume this is true also for other relkinds? I have >1G indexes here, but tables are <1G. TIA.
Вложения
Ray Stell <stellr@vt.edu> writes:
> On Apr 28, 2014, at 11:07 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Note however that the issue could only arise on tables that exceed 1GB
> I assume this is true also for other relkinds? I have >1G indexes here, but tables are <1G. TIA.
Right, such indexes would be at risk. However, if it's only the indexes
that exceed 1G, you could consider reindexing those indexes as an
alternative to a full resync.
regards, tom lane