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

[Condor-users] Hetergeneous dagman submit woes submit on windows & process on linux



Hi Kent,

I'm trying to submit dagman jobs from windows machines and have the jobs be processed on linux machines.
I can do this with condor_submit. I'm having trouble getting condor_submit_dag to work in this manner though. I use UNC
paths which mimic the linux paths. I've applied a special windows registry edit to allow UNC paths to be used within the windows shell as well.

Have you been able to do the above? Any special notes or tips?

I notice condor_submit_dag automatically translates
the windows slashes ("\") in to unix slashes ("/").  Since the job is submitted from windows would the unix slashes confuse the
windows condor_submit_dag?

I am able to submit jobs from OS X and have them process on linux using dagman with no problem.



Here is the error I get from the linux starter:

007 (359.000.000) 10/01 02:04:17 Shadow exception!
Error from starter on vm1@xxxxxxxxxxxxxxxxxx: Failed to execute '/usr/local/condor/local.render0001/execute/dir_23605/condor_exec.exe' with arguments 1 1: No such file or directory
0  -  Run Bytes Sent By Job
173  -  Run Bytes Received By Job


Here is the dagman out log..i highlighted interesting areas in red....


10/1 02:02:32 DAGMAN_PENDING_REPORT_INTERVAL setting: 600
10/1 02:02:32 argv[0] == "condor_scheduniv_exec.358.0"
10/1 02:02:32 argv[1] == "-Debug"
10/1 02:02:32 argv[2] == "3"
10/1 02:02:32 argv[3] == "-Lockfile"
10/1 02:02:32 argv[4] == "//condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag.lock"
10/1 02:02:32 argv[5] == "-Condorlog"
(I tried forcing the log to use windows slashes in the next line...)
10/1 02:02:32 argv[6] == "\\condormaster\Production\Render_Jobs_Output\maya\linux_ball_10_01_2007_02_01_00\linux_ball_dag.log"
10/1 02:02:32 argv[7] == "-Dag"
10/1 02:02:32 argv[8] == "//condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag"
10/1 02:02:32 argv[9] == "-Rescue"
10/1 02:02:32 argv[10] == "//condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag.rescue"
10/1 02:02:32 argv[11] == "-MaxIdle"
10/1 02:02:32 argv[12] == "30"
10/1 02:02:32 argv[13] == "-MaxJobs"
10/1 02:02:32 argv[14] == "100"
10/1 02:02:32 argv[15] == "-MaxPre"
10/1 02:02:32 argv[16] == "1"
10/1 02:02:32 argv[17] == "-MaxPost"
10/1 02:02:32 argv[18] == "1"
10/1 02:02:32 DAG Lockfile will be written to //condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag.lock
10/1 02:02:32 DAG Input file is //condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag
10/1 02:02:32 Rescue DAG will be written to //condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag.rescue
10/1 02:02:32 All DAG node user log files:
10/1 02:02:32   //condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball_SubmitLog.txt (Condor)
10/1 02:02:32 Parsing //condormaster/Production/Render_Jobs_Output/maya/linux_ball_10_01_2007_02_01_00/linux_ball.dag ...
10/1 02:02:32 Dag contains 24 total jobs