Bug #4034
closed
"Invalid memory access" when running workflow from the command line
Added by Chad Berkley over 15 years ago.
Updated over 15 years ago.
Description
When running even very simple workflows from the command line, Kepler fails with the following output. For some reason, I can't reproduce this on windows. The invalid memory access error makes me think maybe the jar file was created on windows and is, for some reason, not valid on osx. Maybe a missing byte or something.
berkley@Slickrock kepler$ ./kepler.sh runwf -gui common/demos/test.xml
args: -runwf -gui common/demos/test.xml
[null] loading: ptolemy.actor.gui.PtExecuteApplication args: [Ljava.lang.String;@480773
[null] 659 ms. Memory: 9076K Free: 2971K (33%)
[null] WARN (org.ecoinformatics.seek.ecogrid.SearchRegistryAction:queryRegistryRewriteConfig:201) Problem looking up registry entries using endpoint:
[null] http://knb.ecoinformatics.org/registry/services/RegistryService
[null] Invalid memory access of location 00000000 eip=90a594a8
Exception in thread "main" Java returned: 138
at org.kepler.build.modules.ModulesTask.execute(ModulesTask.java:88)
at org.kepler.build.runner.Kepler.main(Kepler.java:68)
Caused by: Java returned: 138
at org.apache.tools.ant.taskdefs.Java.execute(Java.java:108)
at org.kepler.build.Run.runSuite(Run.java:189)
at org.kepler.build.Run.run(Run.java:145)
at org.kepler.build.modules.ModulesTask.execute(ModulesTask.java:84)
... 1 more
-- Nested Exception ---
Java returned: 138
at org.apache.tools.ant.taskdefs.Java.execute(Java.java:108)
at org.kepler.build.Run.runSuite(Run.java:189)
at org.kepler.build.Run.run(Run.java:145)
at org.kepler.build.modules.ModulesTask.execute(ModulesTask.java:84)
at org.kepler.build.runner.Kepler.main(Kepler.java:68)
berkley@Slickrock kepler$
reverting to an older org.ecoinformatics.ecogrid.RegistryService-stub.jar version did not work (still had the same error)
removing the call to the EcogridService does not fix the invalid memory error. This doesn't have to do with that service/jar - it's a much "larger" issue.
We've got a few layers here: java calling ant that calls java...
any ideas on where to begin?
This same error occurs when we use the "run-workflow" ant target:
$> ant run-workflow -Dworkflow=../common/demos/getting-started/04-HelloWorld.xml
run-workflow:
[run] args: -runwf /Users/leinfelder/workspace_kepler/kepler/build-area/../common/demos/getting-started/04-HelloWorld.xml
[run] 286 ms. Memory: 9780K Free: 3942K (40%)
[run] WARN (org.ecoinformatics.seek.ecogrid.SearchRegistryAction:queryRegistryRewriteConfig:201) Problem looking up registry entries using endpoint:
[run] http://knb.ecoinformatics.org/registry/services/RegistryService
[run] Invalid memory access of location 00000000 eip=91af94a8
re-assigning to David - maybe he can shed some light on what might be happening.
(Unfortunately it seems like it might be isolated to Macs?)
Does kepler.sh even exist anymore? I am not finding it in any of the modules I have downloaded. I cannot reproduce this problem using the 'ant run-workflow' command anymore, although I did used to get this error intermittently. Unless someone confirms that this error is still occurring for them, I am going to close this bug shortly.
Run the 'ant starupt-script' command to build the kepler.sh file. Not sure if this is still a bug with the new script or not.
This bug is no longer occurring on the Mac with either kepler.sh or with 'ant run-workflow.' However, kepler.sh does throw an exception when you choose to exit using ctrl-c, and that is not aesthetically pleasing. However, this bug no longer exists. I am closing it.
Original Bugzilla ID was 4034
Also available in: Atom
PDF