Add ntasks and cpus_per_task options, fix db_conn in slurmdb_jobs #146
Conversation
pyslurm/pyslurm.pyx
Outdated
|
|
||
| def __dealloc__(self): | ||
| pass | ||
| self.__free() |
There was a problem hiding this comment.
Since __free() is "private" and not used elsewhere, I think we can move the xfree and slurmdb_connection_close directly under __dealloc__.
There was a problem hiding this comment.
no problem. i intended to follow the structure of some of the other classes.
|
Thank you for this contribution! I do need to drop support for Python 2.6. I tried about year ago or so and some folks asked me to hold off, but I think it's time. Also, Python 2.7 support is going away on Jan 1 and may projects have pledged to drop support for Python 2. I'm considering the same for PySlurm. Python 3 support has been working for over a year now. I want to drop support for that CentOS 6 container altogether. |
| # desc.min_cpus = job_opts.get("ntasks") | ||
|
|
||
| # TODO: ntasks_set, cpus_set | ||
| if job_opts.get("overcommit"): |
There was a problem hiding this comment.
Could we work on some tests for these changes?
There was a problem hiding this comment.
submit_batch_job does not seem to be tested beyond giving a wrap script and a job_name. I added simple tests for ntasks and cpus_per_task...
| test_job_id = pyslurm.job().submit_batch_job(test_job) | ||
| test_job_search = pyslurm.job().find(name="name", val="pyslurm_test_job") | ||
| assert_true(test_job_id in test_job_search) | ||
| assert_equals(test_job_search["cpus_per_task"], 3) |
2 participants
This PR builds on top of #141
It includes:
submit_batch_job: "ntasks", "cpus_per_task", "min_cpus", "max_cpus", "min_nodes", max_nodes":submit_batch_jobmethod is along the lines of the_fill_job_desc_from_optsfunction insrc/sbatch/sbatch.cof the Slurm source. Accordingly I adjusted the options involving "ntasks_set" and "cpus_set" which are both pretty much set along with "ntasks" and "cpus_per_task" in the Slurm source.slurmdb_jobs:db_connand close when theslurmdb_jobsinstance is freed.job_condwhich has been allocated in__cinit__.@giovtorres Regarding the Python 2.6 tests: These seem to fail because of a missing "debug" partition in the CentOS 6 docker containers (newer Python versions use CentOS 7). Couldn't you just drop official Python 2.6 support?