On Monday 27 August 2007 17:29, David Rees wrote:
> On 8/27/07, Dan Falconer <lists_slony1-general at avsupport.com> wrote:
> > After grepping through the logs, I've noticed that the time it takes to
> > complete each sync grows at a fairly regular rate.  I'm not sure what the
> > problem is... here's a couple of logs, and sorry in advance for the size.
>
> If you can get these posted up somewhere where the rest of the slony
> guys can see it, they may have more insight, I'm still relatively new
> to slony myself. :-) I'll browse through the logs as soon as I can.

The logs for my slave database server are available at 
http://www.crazedsanity.com/downloads/_slony_slave.tgz for a while.  If you 
are so inclined, please download and peruse them.  It's about a 2 meg 
download.

Slony is now 6+ days behind on the slave server.  I just upped the max sync 
group size from 25 to 500.  I am hoping that has some significant impact, as 
soon I'm going to have to go through the process of dropping schema and 
re-initializing the whole thing.

For what it's worth, it seems to me that the failed log switch has a lot to do 
with the problem.  I mean, Slony is retrieving all of the events, as the 
slave's log files are fairly massive (sl_log_1=413981, sl_log_2=3935888)... 
or something.  I'm very confused as to where things are breaking down, and I 
just don't know what to look for anymore.  SYNCs are processing, seeming to 
indicate that it's TRYING to work... if somebody could shed light on what to 
look for in the logs, that would be great.  I ran the 
test_slony_state-dbi.pl, which started to give interesting information, but 
suddenly stopped with an error "sh: - : invalid option" when trying to 
show "old open connections".
>
> -Dave

-- 
Best Regards,


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