fathom adminserver java heap problem

por | 20 diciembre, 2016


The reason why the adminserver starts when disabling fathom, is because this machine has so many CPU’s
There’s a bug in the 3rd Party Orient database, which means that we need to upgrade it to get the fix.

I now understand why they may have added the parameter:
-XX:+UseParallelGC -XX:ParallelGCThreads=4

However their heap size is still rather small, and they have also added this to the global plugin (which they shouldn’t)

Could you please make the following configuration changes:

1. Add the property «envinronment.concurrency.level=8» to DLC/fathom.init.params
(Note: the word envinronment is misspelled as this is how the third party developers coded it in the oglobal configuration):

envinronment.concurrency.level=8 <- here

2. In the AdminServerPlugins.properties

a. [PluginPolicy]
FROM: jvmargs=-Xmx256m -Xms128m -XX:+UseParallelGC -XX:ParallelGCThreads=4
TO: jvmargs=

b. [PluginPolicy.Progress.AdminServer]
FROM: jvmargs=-Xmx256m -Xms128m -Djava.awt.headless=true -Dsun.lang.ClassLoader.allowArraySyntax=true -XX:+UseParallelGC -XX:ParallelGCThreads=4
TO: jvmargs=-Xms256m -Xmx512m -XX:MaxPermSize=128m -Djava.awt.headless=true -Dsun.lang.ClassLoader.allowArraySyntax=true -XX:+UseParallelGC -XX:ParallelGCThreads=5

Since you’ve hit the Java OOM condition:

Stop the AdminServer on the Fathom machine.
Ensure that there are no «java.exe» processes still running, if there are then kill them because these will be orphan processes due to java running out of memory. Alternatively, reboot the machine.
Restart the AdminServer and fathom
Access the OE Console

Básicamente es aumentar el tamaño de varios parámetros. En el caso de –Xmx y –Xms propone mas chicos, pero ya tuve un caso en el que no funciono

hasta que se puso grande.

Hay que probar en Solaris 11 si todo ya funciona bien.