Details
-
Type:
Bug
-
Status:
Resolved
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: JRuby 1.6.7
-
Fix Version/s: JRuby 1.7.0.pre1
-
Component/s: Core Classes/Modules, Windows
-
Labels:
-
Environment:32 bit windows XP environment with maxed out ram
jruby -v
jruby 1.6.7.2 (ruby-1.8.7-p357) (2012-05-01 26e08ba) (Java HotSpot(TM) Client VM 1.6.0_30) [Windows XP-x86-java]
-
Number of attachments :
Description
When using a simple TCP/IP server derived from ruby's standard GServer jruby will run out of memory. MRI ruby will not. I could not find any combination of the following java arguments to alleviate the problem:
-XX:MaxPermSize=256M -XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:+CMSClassUnloadingEnabled -XX:+HeapDumpOnOutOfMemoryError -Xmx512m -Xms256m
on my box the +HeapDumpOnOutOfMemoryError will motivate the generation of the hprof file in 6-8 minutes every time. I moved the code to an 8 gig 64 bit windows 7 box. The issue still occurs but takes longer (I do increase the memory amounts).
Launching via jruby or 'java -server -jar .\lib\jars\jruby-complete-1.6.7.2.jar' makes no difference. Only launching via mri ruby resolves the issue.
The issue was originally detected in 1.6.5 and I upgraded jruby to see if the issue went away (It did not).
To see the issue launch as follows:
jruby mem_test.rb
then launch
jruby mem_test_client.rb
I am desperate for a fix as we are trying to port our MRI based rails 2.3.x environment to jruby/Rails 3.2.3 and we have an internal tcp/ip server causing total lockup.
Thanks!
Cris
I tested this on my Mac OS X laptop.