daleman

Performance Issues caused by Discovery_Server Probe

Discussion created by daleman on Oct 25, 2013
Latest reply on Oct 28, 2013 by daleman

I'm running NMS 6.5 and have been experiencing some major performance issues when using Infrastructure Manager logged into the Primary Hub.  We do not experience any issues while logged into any other hub.  I currently have a case open with Nimsoft and we have found the issue resides with the discovery_server probe.  If I disable this probe the performance issues are resolved.

 

We have upped the memory on the probe which seemed to have improved a bit but still no where near as good as it should be.  The default memory settings for this probe are mem init - 32m and mem max 1024.  We have set it as high as mem init - 1024 and mem max - 4096m.  This is obviously not needed as I have checked the java process and this probe has only been using as much as 800mb.

 

Our Primary Hub server is on a virtual machine running Windows 2008R2 with 16Gb of memory, with 4 virtual cpus.  We do not run Infrastructure Manager directly on this machine but instead run it off multiple other 2008 servers with plenty of memory (16GB) and cpu.

 

Some of the performance issues we are having are: clicking on hub to expand it seems to stall and sometimes freeze, deploying probes takes a long time and sometimes will freeze, opening up probes to edit will take longer than usual, searching for a robot takes on average over 1 minute. (only takes 3-4 seconds at most with the other hubs)

 

 

We first experienced this about 2 weeks ago when we migrated all of our robots from another domain (NMS 6.5) to our current domain.  We didn't experience the issues in our other domain but also noticed the discovery server was not even running then.

 

Has anyone else experienced this or have  a fix as this is a big issue for us that we have been struggling with over 2 weeks now?

Outcomes