Horton

Re: [BUGS] Concurrent ALTER SEQUENCE RESTART Regression

Peter Eisentraut <peter.eisentraut@2ndquadrant.com>

View thread View on postgr.es
From: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
To: Andres Freund <andres@anarazel.de>, Tom Lane <tgl@sss.pgh.pa.us>
Cc: Petr Jelinek <petr.jelinek@2ndquadrant.com>, pgsql-hackers@postgresql.org, Noah Misch <noah@leadboat.com>, Jason Petersen <jason@citusdata.com>, Michael Paquier <michael.paquier@gmail.com>
Date: 2017-05-12T14:28:10Z
Lists: pgsql-bugs, pgsql-hackers 
On 5/11/17 17:28, Andres Freund wrote:
> Isn't that pretty much the point?  The whole open_share_lock()
> optimization looks like it really only can make a difference with
> subtransactions?

Yeah that confused me too.  That keep-the-lock-for-the-whole-transaction
logic was introduced in a2597ef17958e75e7ba26507dc407249cc9e7134 around
7.3, way before subtransactions (8.0).  The point there was apparently
just to avoid hitting the lock manager on subsequent calls.  That code
has since been moved around end refactored many times.  When
subtransactions were introduced, the current logic of keeping the lock
across subtransactions was added in order to keep the original intent.

-- 
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Make ALTER SEQUENCE, including RESTART, fully transactional.

  2. Modify sequence catalog tuple before invoking post alter hook.

  3. Use weaker locks when updating pg_subscription_rel

  4. Add pg_sequence system catalog

  5. Modify sequence state storage to eliminate dangling-pointer problem