Running CLM server 6.0.2 (RTC only) on Amazon Cloud linux
I've installed RTC 6.0.2 on an Amazon Cloud instance. I downloaded the plain zip with the entire RTC server. Just unzipped and executed server/server.startup.
Before starting i needed to set hard/soft nofile to 65536 and do an "unset JAVA_HOME" to remove references to the OS java installatio (which is incompatible).
It seems to be running fine, but after around 3 days of running, the server suddenly stop. Usually with an error like this in ccm.log:
2016-07-30 01:10:13,405 [ccm: AsynchronousTaskRunner-2 @@ 01:04] WARN .ibm.team.jfs.app.discovery.utils.RootServicesUtil - CRJZS0434W The root services document could not be fetched after 3 attempts.
Then i start the server again and it runs fine again for at few days. I'm thinking this might be because I started the server as a normal user and not root? And the job is ended by the OS or something?
I'm now testing starting the job with "sudo" but I don't have really high hopes for this working.
Any thoughts?
|
Accepted answer
I would look at the was logs for things like OutOfMemoryException as well as possible javacore, heapdump files in the was profile directory.
Morten Madsen selected this answer as the correct answer
Comments
Morten Madsen
commented Aug 04 '16, 7:34 a.m.
That's a very good point. I am running RTC on an absolute minimum virtual linux with 2GB of memory. I'd edited the server.xml only to have 2GB settings, but now i'm trying with 1600MB to see if that will help.
I guess it will help if the OS ran out of memory because java VM pre-allocated too much memory. But if the java VM ran out, then this will definitely not help.
sam detweiler
commented Aug 04 '16, 7:52 a.m.
look in the log files, don't guess.. waste of your time guessing over and over..
Minimum - absolute minimum memory requirement is 8GB.
Morten Madsen
commented Aug 04 '16, 9:45 a.m.
Yeah, I know I'm squeezing it a bit here ;-). There's nothing helpful in the logs I'm afraid.
I'll just see what happens now.
On the larger amazon linux with 4GB memory I ran for a couple of weeks without incidents with this setup.
ps. I think the IPhone 6s has 2 GB of ram :-P.
|
2 other answers
The server has run for 4 days without incident now. It seems like it was some kind of memory error, but probably the OS maxing out on memory, because I set java to grap all 2GB.
With a setting in server.xml of 1600MB it seems to work. Thanks for all the sparring guys.
## EDIT ##
For documentation purposes I will post my findings:
I ended up running with a memory setting of 1024 MB for the RTC application. It seems like the most important thing is that the operating system has enough memory, or it will kill the rtc java process. RTC has been running with 1024MB and avg. 36% free for 2 weeks now.
I know this configuration is not supported, but Amazon cloud gets SO much more expensive, if you want to have 4GB of memory on your box, and my solution seems to work fine. Also we are uploading big files (200MB+) to the SCM because we're doing game development with Unreal engine and have very big assets. RTC seems to handle all this perfectly :). I really don't know why anyone in the game industry would select Perforce instead of RTC, which seems to be a case of legacy, and the main choice for game dev.
|
is the app server up, but jazz is down, or is the app server down
Comments
Morten Madsen
commented Aug 01 '16, 12:49 p.m.
The app server (WAS liberty profile) is down. The entire java process is shut down somehow (can't see it with 'top' or ps -aux anymore). Hmm, starting it up with "sudo" doesn't seem to make a difference. The java process is shut down after 3 days... pretty much on the clock every time. weird. |
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.