Z3 support - Artifact test

[roydahan]

Testing

• https://argus.scylladb.com/tests/scylla-cluster-tests/92fb8a0b-fef4-4215-8450-19340746ed7b
• Run 8 - final test with the ssd treatment and 22 cores
• Run 7 - test with 44 cores

PR pre-checks (self review)

• I added the relevant backport labels
• I didn't leave commented-out/debugging code

Reminders

• Add New configuration option and document them (in sdcm/sct_config.py)
• Add unit tests to cover my changes (under unit-test/ folder)
• Update the Readme/doc folder relevant to this change (if needed)

[roydahan]

Since I want to add the z3 instance types to the same job with other instance types (n2), I need to think of a way to change this one.

[roydahan]

@soyacz maybe I can add it to the "if" in gce_utils?

[fruch]

In a general, we could think of overrides per instance types, if we still want upfront configuration to be correct.

Other option is to clear this option in configuration by default, and have code that selects it automatically based on known types, and other tests should set it as needed.

[soyacz]

people will make mistakes (or will cause conflicts like this) and each new test that will use z3's will be failing because of trying to attach disks.
We could think of overrides per instance type (as we might want to add fewer disks to smaller n2 instances) - but I think it's beyond scope of this PR and should be added as follow up.
For this commit, I'd add to already added special handling for Z3's. Inside of if "z3-highmem" in machine_type: clause i'd add: instance.disks = [disk for disk in disks if "-data-local-ssd-" not in disk.device_name] - this should remove local ssd's from these types and workaround the issue.

[fruch]

So would suggest dropping the default for it, and putting it in the code like that

import re

# mapping between GCE instance types and their number of local SSDs disk we can use for data storage
gce_instances = {
    "n2-highmem-8": 2,
    "n2-highmem-16": 4,
    "n2-highmem-32": 8,
    "n2-highmem-48": 16,
    "n2-highmem-64": 24,
    "n2-highmem-80": 24,
    "n2-highmem-96": 24,
    "n2-highmem-128": 24,
    "z3-.*": 0,
    "c3-.*": 0,
    "c4-.*": 0,
}


def test_01():
    list_gcp_instances = [
        "n2-highmem-8",
        "n2-highmem-16",
        "n2-highmem-32",
        "n2-highmem-48",
        "n2-highmem-64",
        "n2-highmem-80",
        "n2-highmem-96",
        "n2-highmem-128",
        "z3-highssd-16",
        "c3-highmem-16",
        "c4-highmem-16",
    ]
    for i in list_gcp_instances:
        for instance_type_regex, local_ssd_count in gce_instances.items():
            regex = re.compile(instance_type_regex)
            if regex.match(i):
                print(f"Instance Type: {i}, Local SSD Count: {local_ssd_count}")

test should still be able to set it base on their need of disks, but if you didn't each type would have default

[roydahan]

I prefer the solution of @soyacz since I don't want to start explicitly listing all types and ssd configurations and we already have a special treatment for Z3 machine type.

[fruch]

you don't have to list them all, you can still have default, but since this default isn't going to be correct for some of the type, I just suggest to organize it in a way it would be clearer, and not build some specific code for each new supported type that should have different value

either way you'll need to adapt the code on each new type that doesn't fit with the defaults.

[roydahan]

Ready for review and passed testing.

[fruch]

LGTM

[fruch]

I hope you read carefully the docs :)

[soyacz]

LGTM