TeamCity 5, Linux Server

Hi there,

I recently had a new TeamCity 5.0 server brought online, running on Linux. It is connected to LDAP and has a MySQL database. Initially the server appeared to work correctly, but got rapidly slower and slower. It has now got to the stage that it is running so slowly that the webpage simply times out.

Our IT admin has taken a look at the box, but he has reported there is stacks of HD/memory/CPU free, so does not know what else he can do.

Does anyone have any suggestions?

Steve

4 comments

First of all check how much memory is available to Java process (-Xmx parameter). Try to increase this value.
If it does not help try to take thread dumps and attach them to this thread: http://confluence.jetbrains.net/display/TCD5/Reporting+Issues#ReportingIssues-HangsandThreadDumps

0

Hi, we have increased the memory available to the machine from 2GB to 4GB, and changed the Java memory to 512m - but it still seems to be grinding to a halt.

Could you suggest what specs we should be looking at for a server with about 25 projects, 10 build agents, and 60 users connecting to it (on and off)?

Also, could you suggest what we should be setting this Xmx value to?

Thanks,

Steve

0

What type of JDK you are using? If it's 64-bit JDK then because of JVM internal specifics (all object pointers are doubled in size), applications starting under the 64-bit JDK require at least 1.5 times more memory. For 32-bit JDK -Xmx512m seems to be reasonable.

If server is slow, try to take several thread dumps: http://confluence.jetbrains.net/display/TCD5/Reporting+Issues#ReportingIssues-HangsandThreadDumps
and attach them here.

0

It's a 64bit box, and until recently, it didn't have the 32bit libraries installed, so it must be a 64bit version of Java. I realised that I needed to use the MaxPermSize override in our case, so have been running with the following for a couple of days now:

JAVA_OPTS="-Djava.awt.headless=true -Xmx2048m -XX:MaxPermSize=512m"


Which is in /TeamCity/bin/catalina.sh

So far we haven't had any more lock ups, although we currently only have one project active on the new server (currently the AccuRev plug-in is not functioning fully with this new setup, so we can't migrate anything over until that is resolved).

Thanks for your help, hopefully this server will be okay now.

Steve
0

Please sign in to leave a comment.