Обсуждение: WAL files in 7.2 vs. 7.1

We have noticed a moderate degradation of performance in our system after
upgrading from 7.1 to 7.2.  We are attempting to narrow the list of
culprits (it could be different sizes of critical tables, or any of a
dozen things).

Where the 7.1 installation reports this in the postmaster log:

DEBUG:  MoveOfflineLogs: remove 000000A3000000D5

every 10-20 minutes, the 7.2 installation says

DEBUG:  recycled transaction log file 00000016000000D1

at about the same frequency.  The WAL settings on each database are

wal_files = 2

(and the rest are defaults).  Does 7.2's approach of recycling log files
take less time than 7.1's approach of deleting and re-creating?  How can
I determine if I have enough WAL files allocated?  Should we consider
moving the pg_xlog directory to another physical disk?  (At present it
points to a partition on a RAID-striped group of disks.)


--
Jeff Boes                                      vox 616.226.9550 ext 24
Database Engineer                                     fax 616.349.9076
Nexcerpt, Inc.                                 http://www.nexcerpt.com
           ...Nexcerpt... Extend your Expertise

Jeff Boes <jboes@nexcerpt.com> writes:
> Does 7.2's approach of recycling log files
> take less time than 7.1's approach of deleting and re-creating?

It should.  In any case, if you are only recycling WAL segments once
every 10-20 minutes, any speedup or slowdown in that operation is
quite unlikely to be causing a noticeable overall performance
difference.  I think you should be looking at query plans and so on.

            regards, tom lane