Prev: Autonomous transaction
Next: [HACKERS] Show schema name on REINDEX DATABASE
From: Heikki Linnakangas on 5 Apr 2010 12:01
Simon Riggs wrote:
> I am very disconcerted that there are still no docs whatsoever to
> describe how the server works in these new modes.

I did add a few paragraphs last week, see
http://developer.postgresql.org/pgdocs/postgres/warm-standby.html#STANDBY-SERVER-OPERATION.
It doesn't explicitly talk about that configuration where both
primary_conninfo and restore_command are missing (or misconfigured), but
it does say that it polls pg_xlog. How does that read to you?

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com

--
Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

From: Simon Riggs on 5 Apr 2010 14:29
On Mon, 2010-04-05 at 19:01 +0300, Heikki Linnakangas wrote:
> Simon Riggs wrote:
> > I am very disconcerted that there are still no docs whatsoever to
> > describe how the server works in these new modes.
>
> I did add a few paragraphs last week, see
> http://developer.postgresql.org/pgdocs/postgres/warm-standby.html#STANDBY-SERVER-OPERATION.
> It doesn't explicitly talk about that configuration where both
> primary_conninfo and restore_command are missing (or misconfigured), but
> it does say that it polls pg_xlog. How does that read to you?

It is better, though it might mislead others into thinking I mean
sufficient, which it is not. I do not mean to be rude, but we must be
clear that a major feature requires matching documentation.

Looking through the code some more I note that their are two timing
related parameters that are hardcoded into XLogPageRead(). At the very
least such things should be #defines, though one of them was previously
configurable using pg_standby, so I would like to see them both
accessible to user tuning.

I was also surprised to note that the Startup process is not signaled by
WALReceiver when new WAL is received, so it will continue sleeping even
though it has work to do.

--
Simon Riggs www.2ndQuadrant.com


--
Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

From: Simon Riggs on 5 Apr 2010 14:48
On Mon, 2010-04-05 at 19:29 +0100, Simon Riggs wrote:

> Looking through the code some more I note that their are two timing
> related parameters that are hardcoded into XLogPageRead(). At the very
> least such things should be #defines, though one of them was previously
> configurable using pg_standby, so I would like to see them both
> accessible to user tuning.

....the code says "we want to react quickly when the next WAL record
arrives" and then sleeps for 100ms.

> I was also surprised to note that the Startup process is not signaled by
> WALReceiver when new WAL is received, so it will continue sleeping even
> though it has work to do.

--
Simon Riggs www.2ndQuadrant.com


--
Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

From: Alvaro Herrera on 5 Apr 2010 15:02
Simon Riggs escribi�:
> On Mon, 2010-04-05 at 17:08 +0900, Fujii Masao wrote:
> > On Sat, Apr 3, 2010 at 6:50 AM, Simon Riggs <sriggs(a)postgresql.org> wrote:
> > > Log Message:
> > > -----------
> > > Check compulsory parameters in recovery.conf in standby_mode, per docs.
> >
> > On the recent discussion (*1), some people argued that specifying neither
> > primary_conninfo nor restore_command in the standby mode is not unreasonable,
> > and we reached the consensus that the setting should be allowed. So this
> > commit doesn't reflect the discussion. How about reverting the commit,
> > and restarting the discussion if you have complaint against the consensus?
>
> The attached patch changes the messages and downgrades FATAL to WARNING.
>
> Comments?

Please note that errdetail must be a complete sentence.




--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

--
Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

From: Simon Riggs on 5 Apr 2010 15:42
On Mon, 2010-04-05 at 15:02 -0400, Alvaro Herrera wrote:
> Simon Riggs escribió:
> > On Mon, 2010-04-05 at 17:08 +0900, Fujii Masao wrote:
> > > On Sat, Apr 3, 2010 at 6:50 AM, Simon Riggs <sriggs(a)postgresql.org> wrote:
> > > > Log Message:
> > > > -----------
> > > > Check compulsory parameters in recovery.conf in standby_mode, per docs.
> > >
> > > On the recent discussion (*1), some people argued that specifying neither
> > > primary_conninfo nor restore_command in the standby mode is not unreasonable,
> > > and we reached the consensus that the setting should be allowed. So this
> > > commit doesn't reflect the discussion. How about reverting the commit,
> > > and restarting the discussion if you have complaint against the consensus?
> >
> > The attached patch changes the messages and downgrades FATAL to WARNING.
> >
> > Comments?
>
> Please note that errdetail must be a complete sentence.

Better?

--
Simon Riggs www.2ndQuadrant.com
 |  Next  |  Last
Pages: 1 2
Prev: Autonomous transaction
Next: [HACKERS] Show schema name on REINDEX DATABASE