Mailing List ArchivesPublic Access |
![[Computer Systems Lab]](https://www-auth.cs.wisc.edu/pics/csl_logo.gif){kind=logo}
|
Mailing List ArchivesPublic Access |
|
Eduardo Bach wrote:
No special configuration is necessary. Condor always creates a scratch directory inside of the condor execute directory. If jobs use file-transfer mode (should_transfer_files=yes in the submit file), then they are automatically started with their working directory set to the scratch directory. In all cases, the environment variable _CONDOR_SCRATCH_DIR is also passed to the job.Dear Dan, Thank you very much to the answered. I'm already using tmpwatch with an interval shorter than I like. The problem is that no matter the scratch directories size that I use, they fill very fast in a time less than the acceptable to use tmpwatch. I would like to find amore dynamic way to list the files in use by jobs that are running, so to delete the rest.It seems that these jobs are writing in another folder than _CONDOR_SCRATCH_DIR therefore are not automatically removed, unless the condor is not configured to do so. Where can I find thisconfirugação, if it exists?
Another question: how do I can check the variable _CONDOR_SCRATCH_DIR which was passed by the condor to the job?
Run a job that prints out its environment. --Dan
Thank you again.Eduardo Bach Dan Bradley escreveu:Condor creates a scratch directory for the job inside of the Condor's EXECUTE directory. The location of this directory is passed to the job in an environment variable _CONDOR_SCRATCH_DIR. This scratch directory is automatically cleaned up by Condor.However, there is nothing stopping the job from creating files in other directories that Condor does not automatically clean up. In such cases, you may need to periodically clean out old unneeded files that have accumulated. On unix systems, the 'tmpwatch' tool is useful for this purpose.--Dan Eduardo Bach wrote:Hello to all, First, please forgive my terrible English. I'm starting to use the condor, so be patience. Looking at the size of scratch directory in worker nodes, I realized that some jobs leave the farm without deleting your files in this directory. My first reaction was to try to list the directories used by the current jobs in the scratch, and delete the rest of the files, which must be of jobs' files already completed. Well, with the command condor_q -l job_id I encounter a lot of information, including the variable Iwd, which I think should give some folder in the scratch, but to my surprise this variable points to the home of the job's owner. I even found a reference to the directory scratch with this command, but nothing specific like a directory created just for this job. Then comes my question: is the condor that create this directory to the job? if so how to get this information? Tank's in advanced. Eduardo Bach _______________________________________________ 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-usersThe archives can be found at: https://lists.cs.wisc.edu/archive/condor-users/_______________________________________________ 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-usersThe archives can be found at: https://lists.cs.wisc.edu/archive/condor-users/------------------------------------------------------------------------ _______________________________________________ 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-usersThe archives can be found at: https://lists.cs.wisc.edu/archive/condor-users/