Gabor Paller wrote:
> Hi,
> 
>  
> 
> We have noticed that CDR generation in high-load configurations really
> hinder the system’s performance. Some SipX subsystem unknown to me
> issues a SELECT statement on the CDR table every 10 seconds and as CDRs
> pile up, the system slows down to crawl.
> 
>  
> 
> Are you aware of this problem? Is there any improvement planned in the area?
> 
>  
> 

That SELECT statement is generated by call resolver trying to check for new
call state events (CSE). One select statement every 10 seconds should not
hinder system performance, unless postgres is busy with something else.

You can try couple of things:
- increase the SIP_CALLRESOLVER_CSE_POLLING_INTERVAL in callresolver-config
 (that would make current calls less detailed but historic stats will still
be OK)
- optimize call_state_events in SIPXCDR database (google postgres VACUUM
ANALYZE)


At one point CSEs were written to DB one by one and not in batches.
Writing, as opposed to reading them, was - IMHO - more likely responsible
for poor DB performance. It's folklore not science: Raymond probably has a
more informed opinion on that.

Damian

PS.
Don't edit callresolver-config directly.
Change /etc/sipxecs/sipxcallresolver/callresolver-config.vm and push
profiles for the server.

D.

_______________________________________________
sipx-dev mailing list [email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-dev
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-dev
sipXecs IP PBX -- http://www.sipfoundry.org/

Reply via email to