Oh, I forgot to mention the required info:
	* SLONY-I:  1.2.10
	* POSTGRES: 8.0.13

On Tuesday 04 September 2007 20:53, Dan Falconer wrote:
> 	I have a Master -> Slave setup for Slony-I replication.  For reasons I've
> been working out on another thread on this list, I have recently been
> plagued with problems from our slave server being constantly behind.  Every
> week for the past month, I've had to rebuild the cluster on monday or
> tuesday, after our main inventory processing is done, to get that slave
> server back up to speed.
>
> 	Today, instead of dropping the schema & going through the painful process
> of initializing the cluster, creating the main set (set #1), then
> subscribing the slave, I've decided just to drop the slave node.  After
> doing this, I was surprised to find that Slony didn't automatically
> truncate all the data in sl_log_1 and sl_log_2 (~30.6 million records)... I
> truncated the tables by hand (it only took about 1 second), and then
> started the slons back up.
>
> 	I was then struck by yet another surprise: Slony seems to like keeping all
> it's log data, even when it's the only running node.  I've noticed there
> have been some cleanup processes, but... well, it seems like part of the
> problem I was running into with the slave server being so desperately far
> behind has something to do with an inability to switch log tables:  Take a
> look at my logs (the "NOTICE" lines always seem to be dumped to the console
> when I restart slons, until I open another console)::::
>
> ************************* SNIP *************************
>
> pl=# select NOW() AS timestamp, * from _pl_replication.sl_node; SELECT
> (select NOW()) AS time, (select count(*) FROM _pl_replication.sl_log_1) AS
> count_of_slony_log_1, (select count(*) FROM _pl_replication.sl_log_2) AS
> count_of_slony_log_2;
>           timestamp           | no_id | no_active |
> no_comment         | no_spool
> ------------------------------+-------+-----------+------------------------
>---+---------- 2007-09-04 14:46:56.86454-05 |     2 | t         | Node 2 -
> pl at 192.168.1.201 | f
> (1 row)
>
>              time              | count_of_slony_log_1 |
> count_of_slony_log_2
> -------------------------------+----------------------+--------------------
>-- 2007-09-04 14:46:56.865127-05 |                    0 |               
> 25142 (1 row)
>
> pl=# NOTICE:  Slony-I: cleanupEvent(): Single node - deleting events <
> 73770
>
> pl=# select NOW() AS timestamp, * from _pl_replication.sl_node; SELECT
> (select NOW()) AS time, (select count(*) FROM _pl_replication.sl_log_1) AS
> count_of_slony_log_1, (select count(*) FROM _pl_replication.sl_log_2) AS
> count_of_slony_log_2;
>           timestamp          | no_id | no_active |        no_comment       
>  | no_spool
> -----------------------------+-------+-----------+-------------------------
>--+---------- 2007-09-04 14:52:27.9332-05 |     2 | t         | Node 2 -
> pl at 192.168.1.201 | f
> (1 row)
>
>              time              | count_of_slony_log_1 |
> count_of_slony_log_2
> -------------------------------+----------------------+--------------------
>-- 2007-09-04 14:52:27.933927-05 |                    0 |               
> 39972 (1 row)
>
> pl=# NOTICE:  Slony-I: log switch to sl_log_2 complete - truncate sl_log_1
>
> pl=# select NOW() AS timestamp, * from _pl_replication.sl_node; SELECT
> (select NOW()) AS time, (select count(*) FROM _pl_replication.sl_log_1) AS
> count_of_slony_log_1, (select count(*) FROM _pl_replication.sl_log_2) AS
> count_of_slony_log_2;
>            timestamp           | no_id | no_active |
> no_comment         | no_spool
> -------------------------------+-------+-----------+-----------------------
>----+---------- 2007-09-04 14:52:51.148288-05 |     2 | t         | Node 2 -
> pl at 192.168.1.201 | f
> (1 row)
>
>              time              | count_of_slony_log_1 |
> count_of_slony_log_2
> -------------------------------+----------------------+--------------------
>-- 2007-09-04 14:52:51.149097-05 |                    0 |               
> 40182 (1 row)
> ************************ /SNIP *************************

-- 
Best Regards,


Dan Falconer
"Head Geek", Avsupport, Inc. / Partslogistics.com
http://www.partslogistics.com