I am developing a program which a single complete process can take up hours. I usually run in a remote cluster nodes which use a scheduler (e.g. PBS or SLURM).
In the case:
- I work on new features in branch [test-A]
- I submit a test-run for this [test-A] branch to a remote cluster node
- I git checkout to branch [master] or another [test-X] branches, for working on other features
- I submit other test-runs while I'm in those branches, while then again, doing a lot of git checkout between different branches.
- NOTES: each of running-test can take hours to finish. So the job from [test-A] will still be running while I do those branch-switching and tests with those different branches.
My question:
- How the operating system (e.g. Linux), or the scheduler systems will handle this? Will it care about where the current HEAD pointer is currently on?
- For each submitted running job, will it always be able to correctly 'run'? Even though I switch and working on different branches while the program is running the test from another branch?
There was a similar question recently here, but it's only specific to ruby. How is it in general?
Aucun commentaire:
Enregistrer un commentaire