per-process memory limit

Last week I increased our cache size to 400,000 entries. That, according to the documentation, is the maximum value allowed. Even though elsewhere it says that number should be 3x the number of assets in the largest backingstore. One of our backingstores has about 200,000 assets so I figured I'd max it out.

In our iwtrace.log file these three lines are appearing every 15 seconds
[PHP][Wed Jun 13 09:00:34 2012] The TeamSite server is approaching the per-process memory limit (used = 1425220 KB, proc-mem avail = 42696 KB).
[Wed Jun 13 09:00:34 2012] Please reduce cachesize and restart server to alleviate the problem.
[Wed Jun 13 09:00:34 2012] Server will deactivate all stores if available per-process memory falls below 1467916 KB[/PHP]

Since used + avail = 'dont fall below' limit, am I okay? Should I reduce the cache size by 1, 10, 100, 1000...? As you might imagine, recycling TeamSite is not a simple task.

Find more posts tagged with

Comments

Adam Stoller

It's an interesting question - be curious to know what Support says (I'm assuming you have a case open with them on this too - yes?)

It also sounds like the server itself may not be well configured, or may have too many things running on it or some such.

Bowker

I reduced the cache size to 300,000 ... still receiving the error. Now I'm going to open a ticket.

Luis David Moreno

I found an article in customers autonomy last week searching for troubleshooting, its related to peer-process memory limit, and shows a way to change the limit vaule of peer-process limit, to give some more tolerance before the content stores get frezeed.

The content stores will get deactivated at 1.4GB rather than the actual per-process limit of 2GB because the TeamSite memory monitor threshold is based on a percentage of the total value. The default percentage for shutdown is 70%. Therefore, since the per-process limit on Windows is 2GB, then 70% of that would be 1.4GB. The default value for the warning threshold is 65% - that is when warning messages will start to get logged in iwtrace.log.
These values can be changed, if desired, by adding these settings to the [iwserver] section of iw.cfg and restarting the TeamSite service:

process_maxmem_warn_threshold= somenumber greater than 65

process_maxmem_threshold= somenumber greater than 70
NOTE: These settings should be changed carefully because if they are set too high, the TeamSite service may not be able to gracefully shut down in the event that per-process memory runs too low.

unfortunately i dont know the right values to assign in those settings, so you will have to test and investigate a little further, maybe providing this info to autonomy can be easier for them to find a solution.

I hope it helps Smiley Very Happy

Pawan

I am running into the same problem with cachesize setting of 100,000. Support recommends to lower it down to 50,000. I tried reducing it to 70,000 in vain. It takes about 4-5 days for the error to show up in iwtrace.log and the error keeps repeating until TeamSite service restart.

Just curious to know if you were able to come to a permanent resolution for this. This is killing our production system as frequent restart is not an option for us.

I am on TS7.2.0 RHEL 5.3

Luis David Moreno

We are in the process of moving Teamsite to a Linux OS, we belive this will stop that kind of errors. I will let you know the outcome, cross fingers!

Regards

Luis Moreno

Pawan

We are in the process of moving Teamsite to a Linux OS, we belive this will stop that kind of errors. I will let you know the outcome, cross fingers!

Regards

Luis Moreno

I just edited my previous post and update the environment details. TS 7.2.0 RHEL 5.3

Which version are you moving to?

myth

We are currently experiencing the same issue. Any updates or fixes? Support is not being very supportive. :-(

Bowker

We have received a patch for a memory issue with JBOSS from support. This patch is for 7.2.1 and a separate patch exists for 7.2.1a.

We have not yet implemented this patch, but hope to this weekend.

myth

We are on TS 7.3.1 on windows 2008 server.

Our issue is the one about iwserver per process memory full. It is different from the JBOSS memory issue.

Good luck with the patch application.

kan

Hi,

We are also facing similar memory issue, when multiple performance test runs executed in the environment or number of concurrent users increased. Getting the below message and the server crashes, without a restart the application never works... we are running on Linux machine.. Let me know if already a solution is identified for the issue..

Our cache size is just 30000... we never changed the default value...

The TeamSite server is approaching the per-process memory limit (used = 2939628 KB, proc-mem avail = 835245 KB).
Server will deactivate all stores if available per-process memory falls below 3774873 KB
The TeamSite server is approaching the per-process memory limit (used = 2939628 KB, proc-mem avail = 835245 KB).
Server will deactivate all stores if available per-process memory falls below 3774873 KB
The TeamSite server is approaching the per-process memory limit (used = 2954664 KB, proc-mem avail = 820209 KB).
Server will deactivate all stores if available per-process memory falls below 3774873 KB
The TeamSite server is approaching the per-process memory limit (used = 2963556 KB, proc-mem avail = 811317 KB).
Server will deactivate all stores if available per-process memory falls below 3774873 KB
The TeamSite server is approaching the per-process memory limit (used = 2965604 KB, proc-mem avail = 809269 KB).
The TeamSite server is approaching the per-process memory limit (used = 2963556 KB, proc-mem avail = 811317 KB).
Server will deactivate all stores if available per-process memory falls below 3774873 KB
Server will deactivate all stores if available per-process memory falls below 3774873 KB
The TeamSite server is approaching the per-process memory limit (used = 2965604 KB, proc-mem avail = 809269 KB).
Server will deactivate all stores if available per-process memory falls below 3774873 KB