Horton

Re: Berserk Autovacuum (let's save next Mandrill)

Darafei Komяpa Praliaskouski <me@komzpa.net>

View thread View on postgr.es
From: Darafei "Komяpa" Praliaskouski <me@komzpa.net>
To: Vik Fearing <vik.fearing@2ndquadrant.com>
Cc: PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>
Date: 2019-03-31T10:40:07Z
Lists: pgsql-hackers 
>
>
> > Idea: look not on dead tuples, but on changes, just like ANALYZE does.
> > It's my first patch on Postgres, it's probably all wrong but I hope it
> > helps you get the idea.
>
> This was suggested and rejected years ago:
>
> https://www.postgresql.org/message-id/b970f20f-f096-2d3a-6c6d-ee887bd30cfb@2ndquadrant.fr


Thank you for sharing the link. I've read through the thread and see you
posted two patches, first being similar but different from mine, and second
being about a different matter.

I don't see "rejected" there, just a common distraction of "you should also
consider this" and time-out leading to "returned with feedback" at the end.

Thing is, we have dead large productions and post-mortems now as your patch
wasn't pushed back in 2016, so situation is different. Let's push at least
first of two patches of yours, or mine.

Which one is better and why?

I believe mine, as it just follows a pattern already established and proven
in autoanalyze. If vacuum comes and unable to harvest some dead tuples, it
will come over again in your case, and just sleep until it gets new dead
tuples in mine, which looks better to me - there's no dead loop in case
some dead tuples are stuck forever.
If someone thinks yours is better we may also consider it for autoanalyze?


-- 
Darafei Praliaskouski
Support me: http://patreon.com/komzpa

Commits

  1. Further improve stability fix for partition_aggregate test.

  2. Improve stability fix for partition_aggregate test.

  3. Attempt to stabilize partitionwise_aggregate test

  4. Fix race condition in statext_store().

  5. Attempt to fix unstable regression tests, take 2

  6. Attempt to fix unstable regression tests

  7. Trigger autovacuum based on number of INSERTs

  8. Fix upper limit for vacuum_cleanup_index_scale_factor