Feature: further documentation on xdmod_plugin usage, core_hours and service units

[ds-04]

Description

It would be very useful to have some further information on the xdmod_plugin, core hours (displayed in coldfront on the allocation) and service units.

At the moment im testing Coldfront using two systems (VM1 coldfront+slurm master/client and VM2 xdmod).

• I've run some jobs through a test slurm instance and with the account created by an allocation (and updated within slurm)
• This has been shredded and ingested in xdmod - I can see the graphs
• I'm executing the xdmod_plugin (for total_cpu_hours) but am not seeing anything happening in CF dashboard.
• At the moment the updating of the core_hours shown on an allocation isn't clear to me.
• I cant see anything in the documentation about service units within CF. Is this something sites have developed completely themselves?

Component

Other

Additional information

No response

[ds-04]

For coldfront xdmod_usage to work... e.g. coldfront xdmod_usage -sx -m total_cpu_hours

It appears that in CF django admin a Resource Attribute Type needs added:

Name	Attribute type name
xdmod_resource	Text

Then the Resource (in this case a slurm cluster) needs this Resource attribute type added

Resource attribute type	Value
xdmod_resource	slurm_cluster_as_named_in_slurm

Now the syncer appears to be connecting (in my case had to deal with self signed test cert on the test xdmod instance I'm running - python requests) ... graphs not updated in CF yet, because it can't find the data xdmod end, but working on that

[ds-04]

Further notes:

• Xdmod self signed cert in testing environ needs to be appended into the Coldfront Certifi cert bundle, in Coldfront venv, so the coldfront xdmod_usage syncer can trust the XMOD url
• Need to ensure XDMOD is using the Slurm account - you must use "pi_column": "account_name" in the the resources.json - https://open.xdmod.org/11.0/configuration.html

[ds-04]

Created PR #633 to add a setup section to the xdmod plugin README.md