On Nov 15, 2007, at 10:07 PM, Christopher Browne wrote:

>
> You *might* get a benefit by increasing the parameter
> "SLON_DATA_FETCH_SIZE", presently set to 10, in slon.h, to some
> moderately larger number.
>
Well, in the logs I see lots of  "fetch 100 FROM LOG;" coming out of  
it.  I didn't change the source, so I guess it changed from the 10 to  
100 (it is set to 100 in that header).

> For the most part, if you are finding things "too slow," then I would
> be more inclined to point firstly at the ideas of:
>
I'm just wondering if it is normal for those things - such as the log  
switch, to take so long.
The DB is pretty tuned.  Its got 2x4 core xeons for cpu, plugged into  
an msa70 with 8 15k sas disks in a raid6. IO shouldn't be much of a  
problem.

I guess I'll dig in more and see what else is happening when it tries  
to switch the logs. I'm also not sure how much of htat time is spent  
waiting for locks and whatnot.  like I said, I just noticed the  
queries popping up in the "queries sucking the most time" list.


--
Jeff Trout <jeff at jefftrout.com>
http://www.dellsmartexitin.com/
http://www.stuarthamm.net/