Re: Interruptible sleeps (was Re: CommitFest 2009-07: Yay, Kevin! Thanks, reviewers!)

Tom Lane <tgl@sss.pgh.pa.us>

From: Tom Lane <tgl@sss.pgh.pa.us>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Greg Smith <greg@2ndquadrant.com>, Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>, pgsql-hackers@postgresql.org
Date: 2010-08-28T01:11:53Z
Lists: pgsql-hackers
Robert Haas <robertmhaas@gmail.com> writes:
> On Fri, Aug 27, 2010 at 5:54 PM, Greg Smith <greg@2ndquadrant.com> wrote:
>> The way the background writer wakes up periodically to absorb fsync requests
>> is already way too infrequent on a busy system.

> Maybe instead of a fixed-duration sleep we could wake it up when it
> needs to do something.

*Any* fixed delay is going to be too long for some people and not long
enough for others; and the very same system might fall into both
categories at different times of day.  I don't think "make
bgwriter_delay customizable" is an adequate answer.  We've put up with
that so far because it wasn't possible to do better given the
infrastructure we had for waiting; but if we're going to try to improve
the infrastructure, we should have the ambition of getting rid of this
type of problem.

			regards, tom lane