Hi Ian,
Thanks for the suggestion but we don't think it's that. On the PG box there 
were no logs indicating that the pool had run out, and on the GeoServer side I 
can't seem to hit the pool limit when it's as low as 20 but using 98 threads to 
test a single instance.

We're thinking the behaviour we're seeing seems more like garbage collection 
locking up an set of instances for a while (once its freed up, those instances 
were going better than the not-locked-up ones). Is GC a plausible mechanism?

Cheers,
Jonathan


---- On Thu, 12 May 2016 13:23:02 +0100 Ian Turton<ijtur...@gmail.com> 
wrote ---- 

most likely cause is that you have saturated the postgis connection pool and 
are waiting for earlier requests to complete.

I can't think of anything else obvious that might cause that problem.


Ian 


On 12 May 2016 at 13:08, Jonathan Moules <jonathan-li...@lightpear.com> 
wrote:
Hi List,
    Doing some performance testing (WMS GetMap) on a GeoServer (2.7.2) with 
PostGreSQL (9.5.2) using JMeter, we're getting some odd slowdowns. It seems 
that for a short while (up to a few minutes) GeoServer massively slows down in 
responding to requests; the "max" times shoot up from a few seconds to about 
60-70 seconds! We've had this across several environments with PostGres, but 
the same tests on Oracle have never exhibited this.
    The issue isn't on the PostGres end - the slowest query there during the 
tests takes about 4 seconds, there's no queuing and the resource usage on its 
(separate to GeoServer) box is negligible.
    If I make the same request to GeoServer directly when not part of this 
testing, it only takes a second or two, so there's nothing obviously wrong with 
the requests either. The problem is also proving difficult to replicate 
reliably.
    The connections to PG are via Tomcat using a JNDI pool.
    Has anyone encountered this sort of behaviour before or have any inkling as 
to the cause?
    Thanks,
    Jonathan



------------------------------------------------------------------------------
 Mobile security can be enabling, not merely restricting. Employees who
 bring their own devices (BYOD) to work are irked by the imposition of MDM
 restrictions. Mobile Device Manager Plus allows you to control only the
 apps on BYO-devices by containerizing them, leaving personal data untouched!
 https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
_______________________________________________
 Geoserver-users mailing list
 Geoserver-users@lists.sourceforge.net
 https://lists.sourceforge.net/lists/listinfo/geoserver-users
 





-- 
Ian Turton

 
 




------------------------------------------------------------------------------
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
_______________________________________________
Geoserver-users mailing list
Geoserver-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-users

Reply via email to