Check dask: -N -m send email, and on DIPC

[yongninglei]

dask without progress bar is working well on ips and local BCBL.

Needs to test:

how to change config.yaml so that -N will give cluster job a proper name.

how to change config yaml to let SGE cluster sending email

how to change dask_scheduler.py so that the name of job goes along with index

[yongninglei]

https://jobqueue.dask.org/en/latest/generated/dask_jobqueue.SGECluster.html

[yongninglei]

https://stackoverflow.com/questions/70694825/how-to-change-dask-job-name-to-sgecluster

[yongninglei]

other useful resources https://docs.dask.org/en/latest/deploying-hpc.html https://dask.discourse.group/t/changing-cluster-job-name-from-default-dask-worker/1315/4 (this one is a slurm one)

Other packages that similar to launchcontainer but they are on slurm

https://pypi.org/project/simple-slurm/

https://github.com/facebookincubator/submitit

https://snakemake.readthedocs.io/en/stable/ (seems this one we can use it to build the workflow？)

https://jobqueue.dask.org/en/latest/generated/dask_jobqueue.SLURMCluster.html

https://jobqueue.dask.org/en/latest/generated/dask_jobqueue.SGECluster.html

https://github.uconn.edu/HPC/parallel-slurm

[yongninglei]

Seems this threads gives the solution to proper name the dask workers https://github.com/dask/dask/discussions/7070

[yongninglei]

    n = 100
    params = {
        'cores' : 2,
        'memory' : '1000MB',
        'processes' : 1,
        'walltime' : '180',
        'job_extra' : [
            '--array 0-{0:d}'.format(n - 1), # submit array of sizen
            '--requeue',
        ],
        'env_extra' : [
            'JOB_ID=${SLURM_ARRAY_JOB_ID%;*}_${SLURM_ARRAY_TASK_ID%;*}',
        ],
    }