Update centos distribution to scos as this prevents scos build testing

[HuijingHei]

See openshift/os#901, when running kola test for scos, get error, this is to fix it

$ cosa kola run ostree.basic
kola -p qemu-unpriv --output-dir tmp/kola run ostree.basic
=== RUN   ostree.basic
--- FAIL: ostree.basic (57.14s)
        harness.go:1443: mach.Start() failed: machine "ea43750e-e04b-4632-b966-caff659a5214" failed basic checks: not a supported instance: scos-coreos
FAIL, output in tmp/kola
Error: harness: test suite failed
2022-06-10T09:12:23Z cli: harness: test suite failed

[travier]

Can you make it support all four:

• scos-
• scos-coreos
• rhcos-
• rhcos-coreos
  so that we can pick & choose freely later which one we want to keep?

[HuijingHei]

Can you make it support all four:

• scos-
• scos-coreos
• rhcos-
• rhcos-coreos
  so that we can pick & choose freely later which one we want to keep?

Looks sane, update is done

[cgwalters]

Personally I think we should just delete this distribution checking code, it's IMO an anti-pattern.

That said, where is scos-coreos and rhcos-coreos coming from? That seems redundant as "rhcos" is short for "rhel coreos" so this is like "rhel coreos coreos" right?

[HuijingHei]

That said, where is scos-coreos and rhcos-coreos coming from? That seems redundant as "rhcos" is short for "rhel coreos" so this is like "rhel coreos coreos" right?

Refer to:

coreos-assembler/mantle/platform/platform.go

Line 486 in 75b7978

out, stderr, err := m.SSH(`. /etc/os-release && echo "$ID-$VARIANT_ID"`)

When $ID and $VARIANT_ID are both defined in /etc/os-release, the value will be like scos-coreos (for example in openshift/os#773).

Maybe checking $ID (rhcos or scos) is enough, but for fedora coreos the $ID is fedora which might cause confusion with original fedora

[travier]

I'm looking at making this less redundant but changing the defaults here might have impacts in other places thus it's easier to add VARIANT_ID=coreos than change existing values.