Slony1 PostgreSQL Replication - How to reduce de log messages???

Author

How to reduce de log messages???

Andy

2005-07-21, 7:24 am

I'm running Slony over a Postgresql 8.0.3 server.

My probles is that the log file grows about 3GB/month. How can somehow write the slony logs into another log file, or how can I configurate postgresql that he does not generate all these logs.

I appreciate any help.

The most of the log messages are:
2005-05-29 21:12:45 CEST 4340 - 5904 LOG: duration: 0.207 ms
2005-05-29 21:12:45 CEST 4340 - 5905 LOG: statement: rollback transaction;
2005-05-29 21:12:45 CEST 4340 - 5906 LOG: duration: 0.089 ms
2005-05-29 21:12:46 CEST 4339 - 5905 LOG: statement: start transaction;set transaction isolation level serializable;select last_value from "_replic_loco".sl_action_seq;
2005-05-29 21:12:46 CEST 4339 - 5906 LOG: duration: 0.293 ms
2005-05-29 21:12:46 CEST 4339 - 5907 LOG: statement: rollback transaction;
2005-05-29 21:12:46 CEST 4339 - 5908 LOG: duration: 0.097 ms
2005-05-29 21:12:46 CEST 4340 - 5907 LOG: statement: start transaction;set transaction isolation level serializable;select last_value from "_replic_loco".sl_action_seq;

And... I usually log the statements... so if this is the solution I will take this out.

Best regards,
Andy.

Christopher Browne

2005-07-21, 9:24 am

"Andy" <frum- JGHWHXdXz1VeoWH0uzbU
5w@public.gmane.org> writes:
> I'm running Slony over a Postgresql 8.0.3 server.
>
> =A0
>
> My probles is that the log file grows about 3GB/month. How can
> somehow write the slony logs into another log file, or how can I
> configurate postgresql that he does not generate all these logs.
>
> =A0
>
> I appreciate any help.
>
> =A0
>
> The most of the log messages are:
>
> 2005-05-29 21:12:45 CEST 4340 - 5904 LOG:=A0 duration: 0.207 ms
> 2005-05-29 21:12:45 CEST 4340 - 5905 LOG:=A0 statement: rollback transact=
ion;
> 2005-05-29 21:12:45 CEST 4340 - 5906 LOG:=A0 duration: 0.089 ms
> 2005-05-29 21:12:46 CEST 4339 - 5905 LOG:=A0 statement: start transaction=
;set transaction isolation level serializable;select[
color=darkred]
> last_value from "_replic_loco".sl_action_seq;
> 2005-05-29 21:12:46 CEST 4339 - 5906 LOG:=A0 duration: 0.293 ms
> 2005-05-29 21:12:46 CEST 4339 - 5907 LOG:=A0 statement: rollback transact=[/color]
ion;
> 2005-05-29 21:12:46 CEST 4339 - 5908 LOG:=A0 duration: 0.097 ms
> 2005-05-29 21:12:46 CEST 4340 - 5907 LOG:=A0 statement: start transaction=
;set transaction isolation level serializable;select[
color=darkred]
> last_value from "_replic_loco".sl_action_seq;
>
> =A0
>
> And... I usually log the statements... so if this is the solution I will =[/color]
take this out.

There are a couple of ways to look at this...

1. You could configure Slony-I to SYNC less often which would lead to
fewer of the per-SYNC-related queries. Look at the "-s" option...

That means subscribers will be further behind the origin, which might
or might not be OK from a business perspective...

2. You could configure PostgreSQL to log a bit less verbosely in
general.

On a lot of our systems, we only log queries that take more than
1000ms to run, as that is generally the sort that are interesting from
a tuning perspective. If you did that, that would cut out virtually
all of these queries.
--=20
"cbbrowne","@","ca.afilias.info"
<http://dev6.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)

Darcy Buskermolen

2005-07-21, 11:30 am

On Thursday 21 July 2005 06:34, Christopher Browne wrote:
> "Andy" <frum- JGHWHXdXz1VeoWH0uzbU
5w@public.gmane.org> writes:
89[color=darkred]
97[color=darkred]
>
> There are a couple of ways to look at this...
>
> 1. You could configure Slony-I to SYNC less often which would lead to
> fewer of the per-SYNC-related queries. Look at the "-s" option...
>
> That means subscribers will be further behind the origin, which might
> or might not be OK from a business perspective...
>
> 2. You could configure PostgreSQL to log a bit less verbosely in
> general.
>
> On a lot of our systems, we only log queries that take more than
> 1000ms to run, as that is generally the sort that are interesting from
> a tuning perspective. If you did that, that would cut out virtually
> all of these queries.

Another option is to just change the duration logging just for the slony=20
replication user using the ALTER USER foo SET ..... functionality via SQL.

=2D-=20
Darcy Buskermolen
Wavefire Technologies Corp.

http://www.wavefire.com
ph: 250.717.0200
fx: 250.763.1759

Sponsored Links