Out of memory problems in RTC
![]()
Pavithra Kasturirangan (429●5●60●56)
| asked Oct 12 '10, 6:26 a.m.
edited Feb 22 '14, 7:59 a.m. by Jennifer Cianchetta-Riordan (251●2)
Hi All,
We have RTC application running in a Linux server with 8GB memory and 3Gb allocated to tomcat in the server.startup file. We have not increased the threads .Thats kept to the default value only in the server.startup file. The database is Oracle and is maintained in another server. Jabber is installed in the application server other than RTC. There is no other application/SW that runs in this server. We have about 150+users on an average working in the server and at times it reaches even 300. RTC usage by developers is via windows client. Often, we face "Out of memory problem" problem and RTC goes to unusable state.The java process will be shown as running but RTC logging page will not come up. The only left out option will be to restart RTC then to bring up the service. Can anyone tell me how to improve this state? I do understand that the solution cannot be generic, however, i would like to understand how the allocation of disks and memory is practiced in other places. If only 4GB can only be configured for tomcat, then how to fine tune the current configurations to avoid out of memory issues? Regards, Pavithra |
4 answers
![]()
If you are getting heap dumps, look at them with Heap Analyzer tool.
https://www.ibm.com/developerworks/community/groups/service/html/communityview?communityUuid=4544bafe-c7a2-455f-9d43-eb866ea60091 We have one RTC server that users are facile crashing with their custom BIRT reports. Analysis with the above shows enormous allocations (e.g. 4g of 5g) of data attributable to BIRT. As to your default JVM of 3G, I thought 4G was the recommended minimum. |
![]()
Dear Team,
We are also facing the same issue not every time, but on an average I would say twice in a month or so.
we face "Out of memory problem" problem and RTC goes to unusable state.The java process will be shown as running but RTC logging page will not come up. It throws 404 error.
The only left out option will be to restart RTC then to bring up the service.
We are looking for some solution to this..
RTC Version : 3.0 -- Has this been addressed in the latest releases.
|
![]()
Pavithra, Anoop,
please proceed as follows: 1. Open a defect on jazz.net / a PMR with IBM Rational Support to capture the environment and circumstances and get somebody to look at this for and with you, until resolved. 2. Create a private query against RTC on jazz.net with [fulltext = "out of memory"] and [type = "defect"], displaying the "found in" field, sorting by "found in". You will find a handful of defects for e.g. 3.0.1 which are resolved (fixed). If these match with your environment, conditions, then it is a safe bet you will improve the situation by migrating to a newer version. 3. Please ensure to approach a continuous delivery scenario for consuming newest versions. Upgrade is generally easier for version n to n+1 and becomes increasingly harder as you let versions pass by. For RTC 3.0 you are currently 9 versions behind 4.0.6, counting only the "c" in a.b.c.d. numbering scheme. And 5.0 is looming on the horizon. 4. If for valid business reasons you cannot migrate to a newer version, please make sure you have updated to the latest fix for your version (i.e. RTC 3.0.1 --> 3.0.1.6) 5. Review the Deployment Wiki, e.g. Configuring and Tuning Tomcat or Deployment Monitoring sections for further helpful advide. 6. Another option to consider is to swap Tomcat for WAS (included with RTC), but this step should only be done if working with support you have identified Tomcat as the source of the defect and fixes are not available for the Tomcat version supported by your RTC version. The key is to involve Rational Support as early as you can when things like "out of memory" are in play. That said, all sizing guides on jazz.net indicate your environment as currently described is definitely within the norm of expected and tested scenarios, i.e. it should run smoothly and stable. Anything else is a defect and needs to be looked at. - Arne |
Your answer
Dashboards and work items are no longer publicly available, so some links may be invalid. We now provide similar information through other means. Learn more here.