Possibility of memory leak in SIM ?

[Mikael Rönnbäck]

Just wanted to check if anyone else has encountered this. I'm running SIM 5.0 on Windows 2003 (SP1 with all security patches), and I'm seeing a repeating little issue here.

When SIM service is started everything runs nicely, the CPU usage is some 5-10 % of the available and its rock steady on 800 MB RAM used. Monitored number of devices (servers, ILO boards, clusters & KVM switches) currently is 2277.

SIM continues this way for about 5-7 days, and then it suddenly jumps to 1,5 GB RAM used, and on the same time gets extremely slow.

A quick restart of the services fixes the error for another week, but why and what to do in a more long term way ?

I've been trying to find out if there's any task running and it seems like the daily device identification is running, at least some of the times this happens.

I know SP2 is right around the corner (in fact downloaded it but found out it wouldn't upgrade and I'm not very keen on complete reinstallation (again)) but is this a known issue or has anyone else seen this happen ?

[C0nner]

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=968834

A similar thread.

[Mikael Rönnbäck]

Thanks but not really, since the other thread discusses a possible memory leak in IE, which is the client side part, I'm more interested in the server part, where IE afaik really is not involved in the same way.

[OlivierV]

Hello Mikael.
It could be usefull to exactly know at what time it happens but check if at that moment you don't have one scheduled task running, like software status polling or inventory. I monitor my SIM server using MRTG and this task is very consuming (CPU/RAM). That's why I restart SIM and SQL every night.

[Mikael Rönnbäck]

Well, I cannot specify any specific time of they day, its more like a degradation until it no longer is realistic to wait for a response (meaning if you need to wait more than 5 min for getting a view/query in the browser it is faster to restart the SIM service which takes approx 3 min).

However I do believe that one or more of the tasks discovery, data collection and software version polling may be running, I do not know of a way to get the status properly but these are the ones that seem to bring about the issue.

Just curious, what is MRTG ?

[Bill Slobodnik]

I am monitoring about 1800 devices and have had similar issues, especially after upgrading to 5.0. I have mssql running on the same server and plan on migrating it to a different server in the near future. In the meantime, I set up a schedueld tasks that restarts the HPSIM service daily to keep the memory usage as clean as possible.
Not exactly the answer, but a couple workarounds.

[OlivierV]

MRTG is originaly designed to graph network traffic on routers but in fact it can graph any SNMP value. We use it to grapgh CPU load, network traffic, memory usage, disk space.. on some of our servers
More informations at http://people.ee.ethz.ch/~oetiker/webtools/mrtg/

[Mikael Rönnbäck]

Thank you Bill, great info to have.