andreadotti / geant4-val-azure Goto Github PK

Great project! Thanks! I have noticed that docker image you are referring from README.md is not there: https://hub.docker.com/r/andreadotti/geant4-val/

Does it impact the deployment? I would like to test Geant on Azure at some point.

Set larger default number of events using shipyard configuration (env variable?)

Jobs must have a unique name also if executed on different pools. Use a naming convention for pool and job names, maybe using also docker container tag/name?

Add default retries to jobs.json to retry failed tasks

az-batch status does not work specifying -j

Simplify command line making -j an alias of -r when using submit. In this way all most used commands accept -j to specify job name

Remove ProcessTest files and move them to ProcessTest repository. Make a generic example

Evaluate the use of task_factory mechanism to simplify generation of tasks. It probably requires a complete re-write of the mechanism here, but it is very interesting and could substantially simplify the code.

When issuing a az-batch status -j all ... the script tries to identify a recipe with name all and then issues status of all jobs.

Output of az-batch status jobs .... goes to standard error instead of std output. Fix that

Add a new command that does:

for job in `az-batch status jobs summary.json 2>&1 | awk '{print $5}' | sed 's/job_id=//g'`; do echo -n $job" " ;az-batch status -j $job summary.json ; done

This print status of all jobs

Need to review some commands. What we need is:

• Delete JOB
• Delete/terminate TASKS

Currently it is not clear from az-batch commands.