Re: [Condor-users] Apparently invalid user log file

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

Horvátth Szabolcs wrote:
> Hi,
>
> After not being able to submit a dagman job from windows to linux, I 
> tried using SSH to start the job directly on the scheduler machine.
> This almost works, but the submission of the jobs by dagman fail and I 
> can't find why. This is the only part of the dagman log that looks 
> relevant, but I don't really understand what it means:
>
>
> 03/10/11 17:48:15 FileLock::obtain(2) - @1299775695.807126 lock on 
> /tmp/condorLocks/31/99/338834.lockc now UNLOCKED
> 03/10/11 17:48:15 init: Opening file 
> /U/projects/shots_comp/435/1299582026/logs/_dagLog.log
> 03/10/11 17:48:15 Opening log file #0 
> '/U/projects/shots_comp/435/1299582026/logs/_dagLog.log'(is_lock_cur=false,seek=false,read_header=true) 
>
> 03/10/11 17:48:15 Error, apparently invalid user log file
> 03/10/11 17:48:15 Error, apparently invalid user log file
> 03/10/11 17:48:15 ReadUserLogHeader::Read(): readEvent() failed
> 03/10/11 17:48:15 
> /U/projects/shots_comp/435/1299582026/logs/_dagLog.log: Failed to read 
> file header
> 03/10/11 17:48:15 ReadMultipleUserLogs: added log file 
> /U/projects/shots_comp/435/1299582026/logs/_dagLog.log (21:404744540) to 
> active list
> 03/10/11 17:48:15 Submitting Condor Node Nuke_0559_0562 job(s)...
> 03/10/11 17:48:15 TmpDir(99)::TmpDir()
> 03/10/11 17:48:15 TmpDir(99)::Cd2TmpDir()
> 03/10/11 17:48:15 submitting: condor_submit -a dag_node_name' '=' 
> 'Nuke_0559_0562 -a +DAGManJobId' '=' '75 -a DAGManJobId' '=' '75 -a 
> submit_event_notes' '=' 'DAG' 'Node:' 'Nuke_0559_0562 -a 
> +DAGParentNodeNames' '=' '"" 
> /U/projects/shots_comp/435/1299582026/submitscripts/Nuke_0559_0562.submit
> 03/10/11 17:48:15 failed while reading from pipe.
> 03/10/11 17:48:15 Read so far:
> 03/10/11 17:48:15 ERROR: submit attempt failed
> 03/10/11 17:48:15 submit command was: condor_submit -a dag_node_name' 
> '=' 'Nuke_0559_0562 -a +DAGManJobId' '=' '75 -a DAGManJobId' '=' '75 -a 
> submit_event_notes' '=' 'DAG' 'Node:' 'Nuke_0559_0562 -a 
> +DAGParentNodeNames' '=' '"" 
> /U/projects/shots_comp/435/1299582026/submitscripts/Nuke_0559_0562.submit
>
>
> "Error, apparently invalid user log file"? What can I do to solve the 
> issue or find the root of the problem?
>
> Cheers,
> Szabolcs

Could you let us know the version of Condor you are working with (run 
condor_version where you submitted the job) and/or let us see the 
contents of file /U/projects/shots_comp/435/1299582026/logs/_dagLog.log ?

thanks
Todd