From: | Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | Greg Stark <gsstark(at)mit(dot)edu>, Greg Smith <greg(at)2ndquadrant(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>, Matt Chesler <matt(at)pragmatrading(dot)com> |
Subject: | Re: SR fails to send existing WAL file after off-line copy |
Date: | 2010-11-02 07:28:06 |
Message-ID: | 4CCFBD86.9090607@enterprisedb.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 02.11.2010 00:47, Tom Lane wrote:
> Greg Stark<gsstark(at)mit(dot)edu> writes:
>> On Mon, Nov 1, 2010 at 12:37 AM, Heikki Linnakangas
>> <heikki(dot)linnakangas(at)enterprisedb(dot)com> wrote:
>>> Yes, indeed there is a corner-case bug when you try to stream the very first
>>> WAL segment, with log==seg==0.
>
>> This smells very much like
>> http://article.gmane.org/gmane.comp.db.postgresql.devel.general/137052
>
>> I wonder if there's some defensive programming way to avoid bugs of this sort.
>
> It strikes me that it's not good if there isn't a recognizable "invalid"
> value for WAL locations. These bits of code show that there is reason
> to have one. Maybe we should teach initdb to start the WAL one segment
> later, and then 0/0 *would* mean "invalid", and we could revert these
> other hacks.
Good idea. That can even be back-patched to 9.0, it should have no
effect on existing installations,
--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com
From | Date | Subject | |
---|---|---|---|
Next Message | Heikki Linnakangas | 2010-11-02 09:40:42 | pgsql: Bootstrap WAL to begin at segment logid=0 logseg=1 (000000010000 |
Previous Message | Peter Eisentraut | 2010-11-02 07:13:58 | Re: [PATCH] Custom code int(32|64) => text conversions out of performance reasons |