ZTS: Optimize KSM on Linux and remove it for FreeBSD

[mcmilk]

This patchset does:

Optimize KSM settings for Linux and disable it for FreeBSD

I reemoved the "Fix FreeBSD problem" with #16822 ... this should maybe go into a seperat PR.

System	Before	New	latest
almalinux8	3h 25m 47s	3h 17m 23s	3h 18m 9s
almalinux9	3h 1m 9s	2h 48m 18s	2h 46m 41s
debian11	3h 31m 43s	3h 12m 58s	3h 16m 35s
debian12	2h 58m 27s	2h 33m 33s	2h 38m 44s
fedora40	3h 2m 11s	2h 43m 19s	2h 48m 50s
fedora41	3h 0m 8s	2h 41m 21s	2h 37m 37s
fedora42	2h 55m 35s	2h 41m 40s	2h 41m 44s
freebsd13-4r	2h 15m 55s	1h 43m 33s	1h 43m 22s
freebsd14-2r	2h 16m 32s	1h 44m 40s	1h 46m 48s
freebsd15-0c	3h 33m 55s	2h 20m 53s	2h 16m 30s
ubuntu20	3h 9m 18s	2h 58m 12s	2h 58m 33s
ubuntu22	3h 44m 4s	3h 28m 41s	3h 37m 23s
ubuntu24	3h 6m 6s	2h 50m 37s	2h 47m 55s
total	1d 16h 1m 13s	1d 11h 5m 34s	1d 11h 19m 18s

Motivation and Context

Some ZTS test runs fail because of memory pressure. This PR tries to fix this.

Description

All commits have the needed infos ;-)

How Has This Been Tested?

Tried here and there, until the result was stable and the times went down ;)

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Performance enhancement (non-breaking change which improves efficiency)
• Code cleanup (non-breaking change which makes code smaller or more readable)
• Breaking change (fix or feature that would cause existing functionality to change)
• Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
• Documentation (a change to man pages or other documentation)

Checklist:

• My code follows the OpenZFS code style requirements.
• I have updated the documentation accordingly.
• I have read the contributing document.
• I have added tests to cover my changes.
• I have run the ZFS Test Suite with this change applied.
• All commit messages are properly formatted and contain Signed-off-by.

[tonyhutter]

Free kernel caches on Linux VMs via cron, to get more usable memory on them.

For Linux, you might be better off playing with some of the virtual memory tunables on either the runner or the VMs. For example, have you tried setting vm.min_free_kbytes to higher values?

[mcmilk]

Free kernel caches on Linux VMs via cron, to get more usable memory on them.

For Linux, you might be better off playing with some of the virtual memory tunables on either the runner or the VMs. For example, have you tried setting vm.min_free_kbytes to higher values?

Yes, I am trying to guess - which screw I should turn.... this is a first try.

Edit: Changing vm.min_free_kbytes didn't work out very well ... :(

[amotin]

Could we commit the version updates independently before jumping into optimizations? I'd really like to see FreeBSD 15 to finally coming clean, while the VM management changes look less straightforward.

[mcmilk]

Could we commit the version updates independently before jumping into optimizations? I'd really like to see FreeBSD 15 to finally coming clean, while the VM management changes look less straightforward.

Should I create a seperate PR for this?

[amotin]

Could we commit the version updates independently before jumping into optimizations? I'd really like to see FreeBSD 15 to finally coming clean, while the VM management changes look less straightforward.

Should I create a seperate PR for this?

Yes, please. I'd be happy to merge it once it reasonably pass CI, hopefully in not too long, and then we could return to this one.

[tonyhutter]

Just want to say that shaving 6 hours off the total run time is really impressive 👍