Mailing List Archives

Public Access

 		[Computer Systems Lab]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Java universe on WinXP, misconfigured?

Hi Ben,
I've reduced the submit file to the bare minimum, but I had to leave in the should_transfer_files = YES because I don't have a shared filesystem set up. I still get the same problem, with the same entries in the remote Starter log, except the lines you pointed out now look like this: 

4/15 21:12:28 Output file: C:\Progra~1\Condor\execute\dir_2932\out.txt
4/15 21:12:28 Error file: C:\Progra~1\Condor\execute\dir_2932\err.txt
As far as I know (please correct me if I'm wrong), the _condor_stdout and _condor_stderr format comes from using $(CLUSTER) and $(PROCESS) to distinguish output from multiple copies of a job, something I do regularly in the vanilla universe, without problems. 

Thanks,

Rob

Ben Burnett wrote:

Thanks.  You're right, it does look fairly straightforward.  I've
even run it on my machine and it works just fine. What troubles me is the log file entries your spits out: 


4/15 16:39:25 Output file:
C:\Progra~1\Condor\execute\dir_3248\_condor_stdout
4/15 16:39:25 Error file:
C:\Progra~1\Condor\execute\dir_3248\_condor_stderr


The only place I find mention of these name is in the Grid Manager,
which I'm guessing you are not using, and when error/log file names
are not specified--which is strange, since you do.

Since you know that the application runs on the machine itself,
would it be possible to simplify the submit file to the most basic one possible: 

universe                = java
executable              = JavaTest.class
arguments               = JavaTest
output                  = out.txt
error                   = err.txt
log                     = log.txt
queue
Then clean out the logs, and start Condor fresh. See if that actually works. 

-B
This 
-----Original Message-----
From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-
bounces@xxxxxxxxxxx] On Behalf Of Rob de Graaf
Sent: Tuesday, April 15, 2008 12:45 PM
To: Condor-Users Mail List
Subject: Re: [Condor-users] Java universe on WinXP, misconfigured?

Hi Ben,

My submit file is fairly straightforward:

universe                = java
executable              = JavaTest.class
arguments               = JavaTest
output                  = output/java.$(Cluster).$(Process).out
error                   = error/java.$(Cluster).$(Process).err
log                     = log/java.$(Cluster).$(Process).log

requirements            = (OpSys == "LINUX" || OpSys == "WINNT51")
should_transfer_files   = YES
when_to_transfer_output = ON_EXIT
notification            = never

queue 10

When I submit this, some will start on WinXP nodes, run into this Java
configuration problem, get evicted and restarted on Linux nodes where
they complete normally.

The application is just hello world:

public class JavaTest {
         public static void main( String[] args ) {
                 System.out.println("Moo!");
         }
}

I compiled on a 1.6 JDK with -target 1.5 option to make it work on the
older JRE's we have on the WinXP boxes, and as I said before, it does
run, just not through Condor.

Thanks!

Rob

Ben Burnett wrote:

Hi Rob:

Could you post the submit file you are using.  I've tried moving

Condor

around and playing with the paths, but nothing seems to spit out the

same

error.  Maybe I can reproduce it with you submission file.

Regards,
-B


-----Original Message-----
From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-
bounces@xxxxxxxxxxx] On Behalf Of Rob de Graaf
Sent: Tuesday, April 15, 2008 10:00 AM
To: condor-users@xxxxxxxxxxx
Subject: [Condor-users] Java universe on WinXP, misconfigured?

Hello all,

I'm trying to run a Java job on a mixed WinXP / Linux pool.
Everything's
fine on the Linux nodes, but on the windows nodes, Java universe

seems

broken somehow. I get this in the StarterLog:

4/15 16:39:24 Initialized IO Proxy.
4/15 16:39:24 File transfer completed successfully.
4/15 16:39:25 Job 16.0 set to execute immediately
4/15 16:39:25 Starting a JAVA universe job with ID: 16.0
4/15 16:39:25 JavaProc:

Cmd=C:\PROGRA~1\Java\jre1.5.0_07\bin\java.exe

4/15 16:39:25 JavaProc: Args=-Xmx501m -classpath
C:\Progra~1\Condor/lib;C:\Progra~1\Condor/lib/scimark2lib.jar;.
-Dchirp.config=C:\Progra~1\Condor\execute\dir_3248\chirp.config
CondorJavaWrapper C:\Progra~1\Condor\execute\dir_3248\jvm.start
C:\Progra~1\Condor\execute\dir_3248\jvm.end JavaTest
4/15 16:39:25 Tracking process family by login "condor-reuse-slot1"
4/15 16:39:25 IWD: C:\Progra~1\Condor\execute\dir_3248
4/15 16:39:25 Output file:
C:\Progra~1\Condor\execute\dir_3248\_condor_stdout
4/15 16:39:25 Error file:
C:\Progra~1\Condor\execute\dir_3248\_condor_stderr
4/15 16:39:25 Renice expr "10" evaluated to 10
4/15 16:39:25 About to exec

C:\PROGRA~1\Java\jre1.5.0_07\bin\java.exe

-Xmx501m -classpath
C:\Progra~1\Condor/lib;C:\Progra~1\Condor/lib/scimark2lib.jar;.
-Dchirp.config=C:\Progra~1\Condor\execute\dir_3248\chirp.config
CondorJavaWrapper C:\Progra~1\Condor\execute\dir_3248\jvm.start
C:\Progra~1\Condor\execute\dir_3248\jvm.end JavaTest
4/15 16:39:25 Create_Process succeeded, pid=2388
4/15 16:39:25 Process exited, pid=2388, status=1
4/15 16:39:25 JavaProc: JVM pid 2388 has finished
4/15 16:39:25 JavaProc: JVM exited normally with code 1
4/15 16:39:25 JavaProc: Wrapper did not leave start record.
4/15 16:39:25 JavaProc: I'll assume Java is misconfigured here.
4/15 16:39:25 JavaProc: unlinking
C:\Progra~1\Condor\execute\dir_3248\jvm.start and
C:\Progra~1\Condor\execute\dir_3248\jvm.end
4/15 16:39:26 Got SIGQUIT.  Performing fast shutdown.
4/15 16:39:26 ShutdownFast all jobs.
4/15 16:39:26 **** condor_starter (condor_STARTER) EXITING WITH

STATUS

0

The Java app runs fine on the machine itself, just not through

Condor.

The
path to the JVM is correct and the machine advertises Java

capability,

as
well as the correct version. What is misconfigured? This is Condor
version
7.0.1 on Windows XP.

Thanks,

Rob de Graaf




_______________________________________________
Condor-users mailing list
To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/condor-users
The archives can be found at: https://lists.cs.wisc.edu/archive/condor-users/