Usage:
batch job create
--name=NAME
--cluster=CLUSTER
--script=SCRIPT
--executable=EXECUTABLE
--destination=DESTINATION
--source=SOURCE
[--companion-file=COMPANION_FILE]
[--outfile-name=OUTPUT_FILE_NAME]
[--suffix=SUFFIX]
[--overwrite]
batch job run [--name=NAMES] [--output=OUTPUT]
batch job fetch [--name=NAMES]
batch job remove [--name=NAMES]
batch job clean [--name=NAMES]
batch job set [--name=NAMES] PARAMETER=VALUE
batch job list [--name=NAMES] [--depth=DEPTH]
batch connection_test --job=JOB
batch cluster list [--cluster=CLUSTERS] [--depth=DEPTH]
batch cluster remove [--cluster=CLUSTERS]
batch cluster set [--cluster=CLUSTERS] PARAMETER=VALUE
Arguments:
FILE a file name
INPUT_TYPE tbd
Options:
-f specify the file
--depth=DEPTH [default: 1]
--output=OUTPUT [default: table]
Description:
This command allows to submit batch jobs to queuing systems hosted
in an HBC center as a service directly form your commandline.
We assume that a number of experiments are conducted with possibly
running the script multiple times. Each experiment will save the
batch script in its own folder.
The output of the script can be saved in a destination folder. A virtual
directory is used to coordinate all saved files.
The files can be located due to the use of the virtual directory on
multiple different data or file services
Authentication to the Batch systems is done viw the underlaying HPC
center authentication. We assume that the user has an account to
submit on these systems.
(SSH, 2 factor, XSEDE-account) TBD.
Experiments:
experiments are jobs that can be run multiple times and create input
and output file sin them
cloudmesh:
experiment:
job:
name: {cloudmesh.profile.user.name}-01
directory: ~/experiment/{experiment.job.name}
output: {cloudmesh.experiment.job.name}/output
input: ~/experiment/{experiment.job.name}/input
script: script.sh
source ,,,
destination: {cloudmesh.experiment.job.directory}
- queue associates with server (cloud)
- job could be run on queue and is associated with one or multiple
servers
- experiment is same as job, but gives some facility to run it
multiple times
I do not know what companion file is
Examples:
batch job run [--name=NAMES] [--output=OUTPUT]
runs jobs with the given names
LOTS OF DOCUMENTATION MISSING HERE
[--companion-file=COMPANION_FILE]
[--outfile-name=OUTPUT_FILE_NAME]
[--suffix=SUFFIX] [--overwrite]