Control the collection of lvm grains via config

[agraul]

What does this PR do?

Refactor options to control grain computation into a single disabled_grains list and make lvm aware of it.

lvm grain collection can take a long time on systems with a lot of volumes and volume groups. On one server we measured ~3 minutes, which is way too long for grains.

This change is backwards-compatible, leaving the lvm grain collection enabled by default.

What issues does this PR fix or reference?

Fixes: #63114

Previous Behavior

lvm grains are always loaded

New Behavior

lvm grains are not loaded when disabled in minion configuration

Merge requirements satisfied?

[NOTICE] Bug fixes or features added to Salt require tests.

• Docs: needs changes, e.g. announce the move to disabled_grains
• Changelog - https://docs.saltproject.io/en/master/topics/development/changelog.html
• Tests written/updated

Does this need a grains/lvm.conf specific test, or is this covered by the __virtual__() / Salt loader tests?

Commits signed with GPG?

Yes

Please review Salt's Contributing Guide for best practices.

See GitHub's page on GPG signing for more information about signing commits with GPG.

[agraul]

I did a first implementation of disabled_grains. I'm not totally happy yet because each grain module / function needs to be updated. If we change the structure of disabled_grains to differentiate between grain modules and grain functions, the loader could take care of skipping disabled grains. I'll work on that next.

[agraul]

Given that the schema can have extra options, is it ok to remove the settings from it?

[Ch3LL]

I think its okay to put them on a deprecation path in favor of the new disabled_grains.

Can you make sure you add a comment:
# salt.utils.versions.warn_until(Potassium, "enabled_fqdns_grains is deprecated in favor of.......)
The only reason being is we grep the cod for warn_until and version name each time we work on removing deprecations.

Also add changelog files for each setting you are removing with the filename deprecated stating the details on what users can now use instead.

[Ch3LL]

Looks like htere are some doc and lint test failures. When those are resolved i'll give this a full review. Thanks for implementing this.

[agraul]

I still have to do the requested changes for the deprecation. If possible, it would be great to have a review for the code changes. I'm a bit conflicted on having the grains do the "disabling" in __virtual__ vs putting it in the loader code. Any opinions on what would be a better fit?

[Ch3LL]

ping @s0undt3ch any thoughts here?

I think I tend to agree it would be nice if this logic was in the loader instead of having to add it to each __virtual__ function but wanted to get your opinion.

[dwoz]

Closing this due to inactivity. Anyone should feel free to re-open it if they want to see it through to the end in one release cycle.

[agraul]

Closing this due to inactivity. Anyone should feel free to re-open it if they want to see it through to the end in one release cycle.

I would like to get this in, I think a multiple minutes grain collection time is pretty bad. I still don't know if my current approach is fine by you or if you'd rather do this in the loader.

[dwoz]

Great, and thanks for the reply. Please re-base and fix the conflicts so we can see if all tests pass.

[twangboy]

Please address the conflicts above