pgsql: Rethink checkpointer's fsync-request table representation.

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: pgsql-committers(at)postgresql(dot)org
Subject: pgsql: Rethink checkpointer's fsync-request table representation.
Date: 2012-07-19 23:28:38
Message-ID: E1Ss098-0000hK-JY@gemulon.postgresql.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-committers

Rethink checkpointer's fsync-request table representation.

Instead of having one hash table entry per relation/fork/segment, just have
one per relation, and use bitmapsets to represent which specific segments
need to be fsync'd. This eliminates the need to scan the whole hash table
to implement FORGET_RELATION_FSYNC, which fixes the O(N^2) behavior
recently demonstrated by Jeff Janes for cases involving lots of TRUNCATE or
DROP TABLE operations during a single checkpoint cycle. Per an idea from
Robert Haas.

(FORGET_DATABASE_FSYNC still sucks, but since dropping a database is a
pretty expensive operation anyway, we'll live with that.)

In passing, improve the delayed-unlink code: remove the pass over the list
in mdpreckpt, since it wasn't doing anything for us except supporting a
useless Assert in mdpostckpt, and fix mdpostckpt so that it will absorb
fsync requests every so often when clearing a large backlog of deletion
requests.

Branch
------
REL9_2_STABLE

Details
-------
http://git.postgresql.org/pg/commitdiff/e3981da37a2f071951e54d74d45ad47733cc74a6

Modified Files
--------------
src/backend/storage/smgr/md.c | 437 ++++++++++++++++++++++++-----------------
1 files changed, 256 insertions(+), 181 deletions(-)

Browse pgsql-committers by date

  From Date Subject
Next Message Jan Urbański 2012-07-20 06:59:38 Re: Re: [COMMITTERS] pgsql: Fix mapping of PostgreSQL encodings to Python encodings.
Previous Message Tom Lane 2012-07-19 17:07:56 pgsql: Send only one FORGET_RELATION_FSYNC request when dropping a rela