I'm currently trying to figure out the same problem but on SJSAS 9.1
UR2. I don't think it's GlassFish's fault since it has been running
fine since May until last month. A few things we noticed pointed to a
leak in a JDBC driver and I found a bug ticket for that driver that
talked about a leak.
I would check to find out when the last time you updated your system,
exactly what changes were done, analyze the source code to see if you
introduced a leak, etc.
Add the following JVM options to domain.xml to help diagnose:
-verbose:gc
-XX:+PrintGCDetails
-Xloggc:gclogfile [Replace gclogfile with your local file path]
-Xcheck:jni
On the next OOME:
- look in server.log to find the exact error referring to"Java heap space."
- capture heap dump and analyze it with tool of choice, such as netbeans
heap analyzer
% jmap -dump:format=b,file=<filename> <pid_of_java_process>
and
% java -version
- look at gc logs from new jvm options
Ryan
glassfish_at_javadesktop.org wrote:
> We have a problem with memory on SJSAS PE 8.2 update 4.
> After few days of work the heap space is overflow. In attached screenshot presented how grow up usage of heap space.
> May be anybody know what a problem here or can help me.
>
> After that heap space usage reached 500-510 Mb (we have -Xmx512m option) server works some time still but then throws OutOfMemoryError.
>
> OS: Linux ALT Master 2.4
> JVM: Sun JVM 1.5.0_16
> [Message sent by forum member 'mikamj' (mikamj)]
>
> http://forums.java.net/jive/thread.jspa?messageID=316524
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>
>
>