Re: WAL Rate Limiting

From: Greg Stark <stark(at)mit(dot)edu>
To: Simon Riggs <simon(at)2ndquadrant(dot)com>
Cc: Robert Haas <robertmhaas(at)gmail(dot)com>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Andres Freund <andres(at)2ndquadrant(dot)com>, Heikki Linnakangas <hlinnakangas(at)vmware(dot)com>, Magnus Hagander <magnus(at)hagander(dot)net>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: WAL Rate Limiting
Date: 2014-02-20 14:29:41
Message-ID: CAM-w4HNq8LZFrAqJ1BM5_H2xiADnF+Mnurk8TnDX3rd9hffPSQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Thu, Feb 20, 2014 at 1:45 PM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
> On 19 February 2014 16:04, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>
>> Well, *I* don't think this is ready to go. A WAL rate limit that only
>> limits WAL sometimes still doesn't impress me.
>
> Could you be specific in your criticism? "Sometimes" wouldn't impress
> anybody, but we need to understand whether you are referring to a bug
> or just making a personal assessment of the patch features. Thanks.

I think it's clear what the question here is. The patch represents a
quick easy win for certain cases but isn't the nice broad solution we
would prefer to have. We face this decision a lot and it's a tough
one. On the one hand we want to set a high standard for ourselves and
not load up the Postgres source with
lots of half-measures, on the
other hand we don't want to let "perfect be the enemy of the good" and
have development stall while we refuse easy wins.

I don't think it would be a terrible thing if this patch went in. And
I also don't think it would be a huge loss if it didn't since I think
we will need IO resource management down the road anyways.

I was looking for a few non-controversial patches to work on at least
at the moment to establish some history before I take on more
controversial patches so I don't think committing this is really a
great idea for me right now. If you wanted to make this a more general
system I have some ideas and I think we could do it fairly easily. We
could make CHECK_FOR_INTERRUPTS have a single flag but if it's set
check the "interrupt type" and know that some types can be handled by
simply sleeping and not throwing an error. There are already some
cases that are handled specially so it wouldn't really even change the
code much. I would be happy to help get something like that committed.

--
greg

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Simon Riggs 2014-02-20 14:43:53 Re: WAL Rate Limiting
Previous Message Simon Riggs 2014-02-20 13:45:27 Re: WAL Rate Limiting