Horton

Re: pgsql: Improve autovacuum logging for aggressive and anti-wraparound ru

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
Cc: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>, Kyotaro HORIGUCHI <horiguchi.kyotaro@lab.ntt.co.jp>, sawada.mshk@gmail.com, tgl@sss.pgh.pa.us, sk@zsrv.org, nasbyj@amazon.com, andres@anarazel.de, robertmhaas@gmail.com, pgsql-hackers@lists.postgresql.org
Date: 2019-03-30T01:08:22Z
Lists: pgsql-hackers 
On Fri, Mar 29, 2019 at 11:22:55AM -0300, Alvaro Herrera wrote:
> Yeah, that looks good to me too.  I wonder if we really need it as LOG
> though; we don't say anything for actions unless they take more than the
> min duration, so why say something for a no-op that takes almost no time?
> Maybe make it DEBUG1.

I think that this does not justify a WARNING, as that's harmless for
the user even if we use WARNING for other skips (see
vacuum_is_relation_owner).  However DEBUG1 is also too low in my
opinion as this log can be used as an indicator that autovacuum is too
much aggressive because there are too many workers for example.  I
have seen that matter in some CPU-bound environments.  I won't fight
hard if the consensus is to use DEBUG1 though.  So, more opinions?
Andrew perhaps?

> s/relfroxzenxid/relfrozenxid/

Sure.
--
Michael

Commits

  1. Skip redundant anti-wraparound vacuums

  2. Improve autovacuum logging for aggressive and anti-wraparound runs

  3. In relevant log messages, indicate whether vacuums are aggressive.

  4. Don't vacuum all-frozen pages.