The default max heap size is 1024 MB (on Unix, this is set in JON_SERVER/bin/rhq-server.sh, and on Windows, it's set in RHQ_SERVER/bin/wrapper/rhq-server-wrapper.conf). The first thing I'd try is doubling that and restarting. If you still eventually run out of heap, then you are probably dealing with a memory leak. In that case, you can add the:
option to the java command line (in the same files mentioned above) and then restart the Server. When the Server runs out of heap again, a heap dump file (j
ava_pid<pid>.hprof) will be written to disk. This dump file can be analyzed to figure out where stuff is leaking into the heap.
For a good article on out of memory errors, see:
Are you using / firing many alerts? Or using a large number of agents. Or large numbers of call time data?
And finally: what version is that?
I only define several alerts about ten . There are four agents .
I think there are no large numbers of call time data.
The version is 2.4 .
Thanks . I will add the parameters to analyze the cause.