On Sun, Dec 11, 2005 at 11:50:25AM +0100, hubert depesz lubaczewski wrote:
> 
>    hi
>    i have database of about 40G running under postgresql 8.0.4
>    i added slony 1.1.2 replication for all tables (about 50 tables, in 5
>    schemas).
>    database is quite loaded - both witl a lot of selects, and tens of
>    thousands of updates/inserts daily.
>    everything worked really fine.
>    i made mrtg graph showing replication lag (taken from st_lag_time in
>    sl_status view).
>    during the day the lag was up to 5 seconds, withusually 2-3 peaks of
>    up to 200 seconds every night (big, fast imports take place at this
>    time).
>    then - on friday - replication seemed to stop - or works *very* slow.
>    even:

If replication is slow on all of the nodes, I would check to see if there is
some connection keeping pg_listener from being cleaned out. 

Try running a vacuum verbose on pg_listener on the origin node and see if
there are lots of dead rows that can't be removed. If there are, then you need
to check pg_stat_activity for some long transaction that's causing the
problem.

Also, I am interested in your mrtg graphing of replication lag. It's a project
that I've been meaning to look into. Would you mind sharing your mrtg 
configuration? 

-- 
Tim Goodaire    416-673-4126    tgoodair at ca.afilias.info
Database Administrator, Afilias Canada Corp.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://gborg.postgresql.org/pipermail/slony1-general/attachments/20051212/78a17eec/attachment.bin