On Friday 24 August 2007 11:21, Andrew Sullivan wrote:
>
> > be cool).  The logs had a combined total of 20+ million records (5M in
> > sl_log_1, 15M in sl_log_2)... it was 4 days behind, so I killed
> > replication completely: dropped the replication schema on both servers,
> > re-init, create/subscribe set.
>
> I don't know if that was worth doing.  Often what do _do_ need,
> however, is to kill the slons and VACUUM ANALYSE the log tables.
> you'll get in trouble with long-running slony applier transactions
> otherwise.
>
> And yes, when catching up, you want to set the snapshot size _much_
> larger.

I'm really unclear as to what you mean by "snapshot size".  I've updated the 
maximum SYNC group size from 6 to 25, and (some time ago, apparently) the sync 
interval was changed from 1000 to 5000.  Thusfar, I haven't seen any 
improvements in how far the slave is behind. 

In fact, the most distressing part out of all of this, which could be due to 
some lack of information, is that the master thinks it's out of date with 
itself:::

pl=# select * from _pl_replication.sl_status ;
 st_origin | st_received | st_last_event |      st_last_event_ts      | 
st_last_received |    st_last_received_ts     | st_last_received_event_ts  | 
st_lag_num_events |      st_lag_time
-----------+-------------+---------------+----------------------------+------------------+----------------------------+----------------------------+-------------------+------------------------
         2 |           1 |        158891 | 2007-08-27 15:59:42.075411 |            
73386 | 2007-08-27 15:55:55.426633 | 2007-08-22 04:35:04.110386 |             
85505 | 5 days 11:24:40.374125
(1 row)

>
> A

-- 
Best Regards,


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