Re: [HTCondor-users] DAGMAN_ABORT_ON_SCARY_SUBMIT

["R. Kent Wenger" <wenger@xxxxxxxxxxx>]

On Wed, 7 May 2014, Jiande Wang wrote:

> I submitted several dagman jobs at almost the same time. Each dagman script 
> is like the following:
>
> Job A ......
> ...
>  I got error message in the dagman output file
>
>
> ERROR: node D1: job ID in userlog submit event (9025.0.0) doesn't match ID 
> reported earlier by submit command (9021.0.0)!  Aborting DAG; set 
> DAGMAN_ABORT_ON_SCARY_SUBMIT to false if you are *sure* this shouldn't cause 
> an abort.
>
>
> Is this because condor can not handle several "D1" job at the same time 
> although they belong to different dagman?

Are your D1 jobs from the different DAGs using the same log file?  If so, 
and you're using a version of HTCondor older than 7.9.0, that is your 
problem.

> Any suggestions on this?

Two possibilities:

1) Change your submit files so that no node job user log file is shared 
between more than one DAG that is running at the same time.

2) Upgrade to a post-7.9.0 version of HTCondor -- that will make 
everything work okay as long as your DAG files have different names.

Kent Wenger
CHTC Team