LSF support

[chaselgrove]

This PR adds an LSF submitter.

@kyleam, can you recommend the best way to test this? I'd model off of condor and slurm, but I only see them in test_orchestrators.py (no references to CondorSubmitter or SlurmSubmitter in tests).

[kyleam]

I'd model off of condor and slurm, but I only see them in test_orchestrators.py (no references to CondorSubmitter or SlurmSubmitter in tests).

Yeah, test_orchestrators is currently the only place that exercises the condor and slurm submitters. The class names aren't there, but the submitter name gets passed to the orchestrator class ("slurm", "condor", ...). For example, here's the slurm fixture and one of the slurm tests:

@pytest.fixture(scope="module")
def ssh_slurm():
    skipif.no_ssh()
    skipif.no_slurm()
    from reproman.resource.ssh import SSH
    return SSH("slurm-res", host="slurm")

[...]

@pytest.mark.integration
def test_orc_datalad_slurm(check_orc_datalad, ssh_slurm):
    check_orc_datalad(ssh_slurm, orcs.DataladLocalRunOrchestrator, "slurm")

So, if you wanted to follow the same approach, you could start by s/slurm/lsf/ and then implementing no_lsf in the skip module.

[chaselgrove]

So, if you wanted to follow the same approach, you could start by s/slurm/lsf/ and then implementing no_lsf in the skip module.

Thanks. I've been trying both that and with a local environment, but the tests are failing for unclear reasons. I'll return to this next week.

[codecov[bot]]

Codecov Report

Merging #553 (cdc583c) into master (9bf7580) will decrease coverage by 0.19%. The diff coverage is 43.63%.

@@            Coverage Diff             @@
##           master     #553      +/-   ##
==========================================
- Coverage   89.84%   89.64%   -0.20%     
==========================================
  Files         149      149              
  Lines       12888    12943      +55     
==========================================
+ Hits        11579    11603      +24     
- Misses       1309     1340      +31     

Impacted Files	Coverage Δ
reproman/support/jobs/submitters.py	73.19% <33.33%> (-7.22%)	:arrow_down:
reproman/support/jobs/tests/test_orchestrators.py	94.48% <63.15%> (-1.10%)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 9bf7580...cdc583c. Read the comment docs.

[chaselgrove]

The test works as it should, but:

I don't use skipif because I use the presence of an environment variable to both trigger and configure the test.

Ideally I'd create job_spec_lsf for the LSF test, but I don't see how pytest would let me use that elegantly (without duplicating check_orc_datalad completely, the only change being check_orc_datalad(job_spec, dataset) -> check_orc_datalad_lsf(job_spec_lsf, dataset)). So I patch job_spec in check_orc_datalad (though I suppose this could happen in job_spec). Happy to consider a more elegant option.

[kyleam]

The test works as it should, but:

I don't use skipif because I use the presence of an environment variable to both trigger and configure the test.

Works for me.

Ideally I'd create job_spec_lsf for the LSF test, but I don't see how pytest would let me use that elegantly (without duplicating check_orc_datalad completely, the only change being check_orc_datalad(job_spec, dataset) -> check_orc_datalad_lsf(job_spec_lsf, dataset)). So I patch job_spec in check_orc_datalad (though I suppose this could happen in job_spec). Happy to consider a more elegant option.

Yeah, I think what you landed on is probably the best option.

While I don't know anything about LSF, nothing looks off to me in terms of where the changes are being made or the general content.

Thanks.

[yarikoptic]

Let's proceed!