Re: Massive parallel queue table causes index deterioration, butREINDEX fails with deadlocks. - Mailing list pgsql-performance

From Jeff Janes
Subject Re: Massive parallel queue table causes index deterioration, butREINDEX fails with deadlocks.
Date
Msg-id CAMkU=1z03uq-GHQ+Dgv7ksEVyTA6QzAbSkxVmj3GK-WGK2W=gw@mail.gmail.com
Whole thread Raw
In response to Re: Massive parallel queue table causes index deterioration, butREINDEX fails with deadlocks.  (Gunther <raj@gusw.net>)
Responses Re: Massive parallel queue table causes index deterioration, butREINDEX fails with deadlocks.
Re: Massive parallel queue table causes index deterioration, butREINDEX fails with deadlocks.
List pgsql-performance
On Sun, Feb 24, 2019 at 1:02 PM Gunther <raj@gusw.net> wrote:

Thank you all for responding so far.

David Rowley  and Justin Pryzby suggested things about autovacuum. But I don't think autovacuum has any helpful role here. I am explicitly doing a vacuum on that table. And it doesn't help at all. Almost not at all.

If you do a vacuum verbose, what stats do you get back?  What is the size of the index when the degradation starts to show, and immediately after a successful reindex?

Also, how is JobID assigned?  Are they from a sequence, or some other method where they are always added to the end of the index's keyspace?

When it starts to degrade, what is the EXPLAIN plan for the query?

Cheers,

Jeff

pgsql-performance by date:

Previous
From: Jeff Janes
Date:
Subject: Re: Massive parallel queue table causes index deterioration, but REINDEX fails with deadlocks.
Next
From: Gunther Schadow
Date:
Subject: Re: Massive parallel queue table causes index deterioration, butREINDEX fails with deadlocks.