Hash Indexes Causing Deadlock Notices - Mailing list pgsql-general

From mark
Subject Hash Indexes Causing Deadlock Notices
Date
Msg-id 39EE30B3.9C5DB089@i4free.co.nz
Whole thread Raw
Responses Re: Hash Indexes Causing Deadlock Notices  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Hi,

I have been experimenting with the Apache Benchmarking Tool ( ab ) to
give my demo web site a bit of a hiding. The database backend is
Postgresql 7.1 ( dev from 25/09/00) .

I noticed that a large number of log notice messages announcing a
deadlock being detected together with  transactions being aborted (and
suggesting that I read the lock manual entry).   I am a bit confused
about the deadlock as I use select...for update to lock rows and process
all tables in the same order ( which SHOULD prevent any deadlocks...)

The tables impacted in the benchmark had HASH indexes. ( only = scans
were being performed, it seemed a good idea at the time ..). Recreating
these indexes as BTREE made all the messages go away. ( and produced a
better request/s result ! )

I am not sure if this issue with HASH indexes occurs in the production
sources :-). I will attempt to replicate if anybody is interested.

Mark


pgsql-general by date:

Previous
From: Neil Conway
Date:
Subject: Re: MySQL PASSWORD('x') function workalike
Next
From: Larry Rosenman
Date:
Subject: Re: [HACKERS] pg_connect error