[Bioc-devel] BatchJobs and BiocParallel

[Michel Lang]

> - BatchJobs: is there support, or are there any plans to support the
>   HTCondor infrastructure [1], and will this back-end be supported in
>   BiocParallel?

We don't have access to a system running condor, otherwise we would
have implemented support already. I'd either need a temporary account
on the system or the following information:

* a minimal condor template to run an R script, including
  + job name being set
  + log file defined
  + resources specified (at least walltime and memory)

* command to submit the job description file to the scheduler
  + format of returned batch ids assigned by the scheduler
     (usually whitespace-delimited list in stdout)
  + exit code on success (should be 0)
  + [optional] temporary error codes

* command to list batch ids
  + preferably only jobs of current user
  + exact output format
  + exit code on success (should be 0)
  + [optional] temporary error codes

* command to kill a job given its batch id
  + exit code on success (should be 0)
  + [optional] temporary error codes

For comparison, see the implementations on
<https://github.com/tudo-r/BatchJobs/wiki/Cluster-Functions>.
To get the exit codes in sh/bash/zsh, execute the respective command
and then "echo $?". We use the temporary error codes to trigger a
sleep-retry-loop. Maybe your cluster admins know some or the condor
documentation provides a table of exit codes.