Status: Resolved (View Workflow)
Resolution: Not A Defect
HP DL380 G7
Xeon E5649 x 2
CentOS 6.2 x86_64
Our log file is over 20MB, when using a self defined parser to parsing it, it always shows out of memory exceptions even if I add the -Xmx4096m parameter.
And I tried to split the log files into several small ones but it seems like warnings plugin collect all of then and parse them together. I think at least this procedure can be adjusted to parsing each file one by one and collect parsing result together in order to save memory.
Sorry, I copied the number using my locale settings (German), the dot is used as separator between thousandths 46.776KB = 46,7MB (in German), in English the separators are used in the opposite way.
Can you please check if your Java process fills up the whole memory of 4GByte?. Is your build on a slave? Does your parser really not contain a newline character?
Ah, yes, the build is on a windows slave with 2GB memory. So it doesn't matter how many master's memory left. I tried add these parameters "-Xms512m -Xmx1024m" to the node's "Launch method" option. But it seems the same. I found that the java process cost more than 311MB memory when this job was running through windows task manager. How can I modify the slave's heap space?
I don't know how to change that. How do you connect your slaves? ssh? jnlp?
I've found the way to change slave's java memory heap. Just add "-Xmx512m" to '<arguments>' in jenkins-slave.xml which is located in slave's JENKINSROOT.
And I connect my slave using jnlp.
I mean MB not KB, 46.776KB is OK in my environment too.