From:
Greg Stark <gsstark(at)mit(dot)edu>
To:
Robert Haas <robertmhaas(at)gmail(dot)com>
Cc:
Andres Freund <andres(at)anarazel(dot)de>, pgsql-hackers(at)postgresql(dot)org
Subject:
Re: Re: Faster CREATE DATABASE by delaying fsync (was 8.4.1 ubuntu karmic slow createdb)
Date:
2010-02-14 15:31:58
Message-ID:
407d949e1002140731j85a4f97nc053ec1b0f3cc458@mail.gmail.com (view raw or flat )
Thread:
2009-12-10 20:41:08 from Michael Clemmons <glassresistor(at)gmail(dot)com>
2009-12-10 21:56:59 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-10 22:01:08 from Michael Clemmons <glassresistor(at)gmail(dot)com>
2009-12-10 22:09:03 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-11 01:38:25 from Nikolas Everett <nik9000(at)gmail(dot)com>
2009-12-11 17:58:39 from "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>
2009-12-11 20:43:59 from Nikolas Everett <nik9000(at)gmail(dot)com>
2009-12-11 20:50:10 from "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>
2009-12-11 21:39:34 from Nikolas Everett <nik9000(at)gmail(dot)com>
2009-12-11 21:57:56 from Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>
2009-12-11 22:12:45 from Scott Carey <scott(at)richrelevance(dot)com>
2009-12-11 22:19:05 from Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>
2009-12-11 21:59:43 from Scott Mead <scott(dot)lists(at)enterprisedb(dot)com>
2009-12-11 22:12:47 from Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>
2009-12-13 03:56:42 from Robert Haas <robertmhaas(at)gmail(dot)com>
2009-12-11 22:39:54 from Greg Smith <greg(at)2ndquadrant(dot)com>
2009-12-11 22:52:01 from Michael Clemmons <glassresistor(at)gmail(dot)com>
2009-12-11 23:59:13 from Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>
2009-12-12 00:19:38 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-12 20:36:27 from Michael Clemmons <glassresistor(at)gmail(dot)com>
2009-12-12 20:38:41 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-28 22:54:51 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-28 22:59:43 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 18:30:49 from Andres Freund <andres(at)anarazel(dot)de>
2010-01-18 16:35:59 from Greg Stark <gsstark(at)mit(dot)edu>
2010-01-19 14:52:25 from Greg Stark <gsstark(at)mit(dot)edu>
2010-01-19 14:57:14 from Greg Stark <gsstark(at)mit(dot)edu>
2010-01-20 04:13:03 from Andres Freund <andres(at)anarazel(dot)de>
2010-01-20 05:21:07 from Greg Smith <greg(at)2ndquadrant(dot)com>
2010-01-27 07:21:44 from Greg Smith <greg(at)2ndquadrant(dot)com>
2010-01-19 15:03:16 from Andres Freund <andres(at)anarazel(dot)de>
2010-01-19 15:25:46 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2010-01-29 18:56:23 from Greg Stark <gsstark(at)mit(dot)edu>
2010-02-02 17:36:12 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-02 17:43:15 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-02 17:50:15 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2010-02-02 18:14:40 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-02 18:34:07 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-02 19:06:32 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-02 19:08:12 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-02 19:33:30 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2010-02-02 19:45:46 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-03 11:53:58 from Greg Stark <gsstark(at)mit(dot)edu>
2010-02-03 12:03:04 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-03 13:42:57 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-03 14:19:49 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-06 05:03:30 from Greg Smith <greg(at)2ndquadrant(dot)com>
2010-02-06 12:03:50 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-07 05:13:15 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-07 09:23:14 from Greg Smith <greg(at)2ndquadrant(dot)com>
2010-02-07 16:24:00 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2010-02-07 18:23:10 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-07 18:27:02 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-08 01:31:42 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-08 03:09:01 from Alvaro Herrera <alvherre(at)commandprompt(dot)com>
2010-02-08 04:53:23 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-08 07:13:41 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-08 18:34:01 from Greg Stark <gsstark(at)mit(dot)edu>
2010-02-08 19:29:46 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-11 02:27:30 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-12 15:49:16 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-14 14:03:44 from Greg Stark <gsstark(at)mit(dot)edu>
2010-02-14 15:31:58 from Greg Stark <gsstark(at)mit(dot)edu>
2010-02-14 17:11:39 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2010-02-14 17:27:00 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-14 17:37:15 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2010-02-14 20:24:24 from Florian Weimer <fw(at)deneb(dot)enyo(dot)de>
2010-02-14 20:41:02 from Mark Mielke <mark(at)mark(dot)mielke(dot)cc>
2010-02-14 20:49:09 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-15 00:08:10 from Mark Mielke <mark(at)mark(dot)mielke(dot)cc>
2010-02-14 20:57:08 from Robert Haas <robertmhaas(at)gmail(dot)com>
2010-02-14 21:43:23 from Andres Freund <andres(at)anarazel(dot)de>
2010-02-14 23:33:54 from Greg Stark <gsstark(at)mit(dot)edu>
2010-01-20 04:02:17 from Andres Freund <andres(at)anarazel(dot)de>
2010-01-20 04:01:55 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-28 23:06:28 from Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
2009-12-28 23:20:35 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 00:46:21 from Greg Smith <greg(at)2ndquadrant(dot)com>
2009-12-29 02:05:39 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 02:53:12 from Michael Clemmons <glassresistor(at)gmail(dot)com>
2009-12-29 02:55:37 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 03:04:06 from Michael Clemmons <glassresistor(at)gmail(dot)com>
2009-12-29 03:11:14 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 10:48:10 from Greg Stark <gsstark(at)mit(dot)edu>
2009-12-29 11:13:21 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-28 23:31:56 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 00:27:29 from Greg Stark <gsstark(at)mit(dot)edu>
2009-12-29 00:29:34 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 00:30:17 from david(at)lang(dot)hm
2009-12-29 00:43:15 from Andres Freund <andres(at)anarazel(dot)de>
2009-12-29 00:46:26 from david(at)lang(dot)hm
2009-12-28 23:57:42 from Thomas Kellerer <spam_eater(at)gmx(dot)net>
Lists:
pgsql-hackers pgsql-performance
On Sun, Feb 14, 2010 at 2:03 PM, Greg Stark <gsstark(at)mit(dot)edu> wrote:
> On Fri, Feb 12, 2010 at 3:49 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>> Greg Stark, have you managed to get your access issues sorted out? If
>
> Yep, will look at this today.
So I think we have a bigger problem than just copydir.c. It seems to
me we should be fsyncing the table space data directories on every
checkpoint. Otherwise any newly created relations or removed relations
could disappear even though the data in them was fsynced. I'm thinking
I should add an _mdfd_opentblspc(reln) call which returns a file
descriptor for the tablespace and have mdsync() use that to sync the
directory whenever it fsyncs a relation. It would be nice to remember
which tablespaces have been fsynced and only fsync them once though,
that would need another hash table just for tablespaces.
We probably also need to fsync the pg_xlog directory every time we
create or rename an xlog segment.
Are there any other places we do directory operations which we need to
be permanent?
--
greg
In response to
Responses
pgsql-performance by date
Next :From: Reydan CankurDate: 2010-02-14 16:36:24Subject : PostgreSQL on SMP ArchitecturesPrevious :From : Greg StarkDate : 2010-02-14 14:03:44Subject : Re: Re: Faster CREATE DATABASE by delaying fsync (was 8.4.1 ubuntu karmic slow createdb)
pgsql-hackers by date
Next :From: Little, DouglasDate: 2010-02-14 15:32:35Subject : function to display ddlPrevious :From : Simon RiggsDate : 2010-02-14 14:59:22Subject : Optimizing GetConflictingVirtualXIDs()
