Remove unused code to automatically select Slurm reservation

Description

We're no longer using a Slurm reservation for calibration jobs - we have partitions (upex-high & upex-middle) to give calibration priority instead. This removes some unused code for xfel-calibrate to automatically decide whether to use a reservation or not.

We're now using the --slurm-partition option from the webservice, which overrides this anyway, and we're not using the --priority option according to caldeploy configs on max-exfl016.

How Has This Been Tested?

CI should exercise the relevant code paths.

Types of changes

• Removal of feature no longer needed

Checklist:

• My code follows the code style of this project.

Reviewers

@danilevc @ahmedk

changed milestone to %3.5.0

added 75 commits

• bffac8ec...3b86f664 - 74 commits from branch master
• 89971d05 - Remove unused code to automatically select Slurm reservation

Compare with previous version

added Waiting for review label

Apart from possibly cleaning up settings, LGTM

1. slurm_scheduling can be removed as well. I don't see a use for it and it can't (and no need to) be used anyway in production. This point could be unrelated to this MR if you are only removing the reservation.

2. https://git.xfel.eu/detectors/pycalibration/-/blob/master/webservice/webservice.py#L914 isn't this related as well and can be removed?

Beside these two points, LGTM.

1. Not per se, even though xcal currently does not even have the privileges, it might actually be a solution to the contention problem during beamtimes. It's cheap to keep the command line argument around in contrast to the unused logic this MR removes.

2. We're currently getting a priority via the API and with the command line flag still in (see 1.), no need to remove it for now. It's most likely not used anyway by myMDC?

Not per se, even though xcal currently does not even have the privileges, it might actually be a solution to the contention problem during beamtimes. It's cheap to keep the command line argument around in contrast to the unused logic this MR removes.

sure.

We're currently getting a priority via the API and with the command line flag still in (see 1.), no need to remove it for now. It's most likely not used anyway by myMDC?

Yes, it's ok for now to leave the other line, where we are receiving this parameter in the myMDC payload. But, we shouldn't do anything with it anymore. https://git.xfel.eu/detectors/pycalibration/-/blob/master/webservice/webservice.py#L915

On the slurm_scheduling I agree with Philipp - it's minimal extra complexity and it could conceivably be useful.

On "priority", @ahmedk is right, the webservice will have to stop passing it to xfel-calibrate. I've made the change.

added 2 commits

• 4e93002d - Remove some unused settings
• 47db95d9 - Don't pass --priority from webservice to xfel-calibrate

Compare with previous version

merged

mentioned in commit 0e8bfd73

removed Waiting for review label