Add optional availabilityZone.from:Machine to disks.

[garloff]

Previously capo defaulted to creating disks in the failureDomain (availability zone) of the virtual machine that is created. In some clouds this is a requirement for good performance or to be able to attach the disk at all.
Other clouds have a global AZ for disks, so we need a pararmeter to control this.

So we introduce three new boolean variables, all defaulting to false: controlPlaneRootDiskPin
workerRootDiskPin
workerAdditionalBlockDevicesPin
which control whether we set
availabilityZone/from: Machine for the respective disks.

Needs testing!

[garloff]

This should address #262.
Needs testing, so I set to draft.

[garloff]

OK, looks good so far. Use openstack-scs2-1-34-v0-git-f4ddbaf for testing.
Here are two clusters with 1+3 and 3+3 and workerRootDiskPin=true and controlPlaneAvailabilityZones=[nbg1,nbg3,nbg6] and controlPlaneOmitAvailabilityZone=false.

linux@infra-mgmt(:openstack):~/scs-training-kaas-scripts [2]$ osc compute server list -f id -f name -f "os-ext-az:availability_zone" -f flavor -f status --sort-key display_name 
┌──────────────────────────────────────┬──────────────────────────────┬────────┬──────────────┬─────────────────────────────┐
│ id                                   ┆ name                         ┆ status ┆ flavor       ┆ OS-EXT-AZ:availability_zone │
╞══════════════════════════════════════╪══════════════════════════════╪════════╪══════════════╪═════════════════════════════╡
│ a2cab588-d258-4429-a780-6aeaee2036ae ┆ zuul-md-2-7jjnp-7sxrb-gqhms  ┆ ACTIVE ┆ SCS-8V-32    ┆ nbg6                        │
│ ff09e6c2-ecc0-491e-806f-a6fc2a4ebc19 ┆ zuul-md-1-vq7j7-psw4m-xtwsk  ┆ ACTIVE ┆ SCS-8V-32    ┆ nbg3                        │
│ ea0d58d4-3215-4489-b74e-bfd0d7d7e86c ┆ zuul-md-0-bnr4n-7jxln-rtgd4  ┆ ACTIVE ┆ SCS-8V-32    ┆ nbg1                        │
│ d978aefb-8990-4d12-9f7c-15268b0e3247 ┆ zuul-4rft9-kvppg             ┆ ACTIVE ┆ SCS-2V-4-20s ┆ nbg1                        │
│ 6a61679c-7cc2-42a6-8259-c39c7369c74a ┆ Infra-Mgmt                   ┆ ACTIVE ┆ SCS-2V-4     ┆ nbg6                        │
│ 29ca3411-e55e-4855-8266-5cbc8c6eba48 ┆ infra-md-2-hxm9s-kdgq7-8hl6d ┆ ACTIVE ┆ SCS-4V-16    ┆ nbg6                        │
│ 021ca9d1-f70e-4f8a-86ff-08d7afcc7239 ┆ infra-md-1-qb4l8-xqck4-4pt2z ┆ ACTIVE ┆ SCS-4V-16    ┆ nbg3                        │
│ ab01ff3d-f865-4e4a-98a4-2fe25314b1b2 ┆ infra-md-0-z5cdp-xf9wc-vfhbt ┆ ACTIVE ┆ SCS-4V-16    ┆ nbg1                        │
│ 5175315d-5429-43ac-965c-0c528fd1e584 ┆ infra-lmj8t-bgl6r            ┆ ACTIVE ┆ SCS-2V-4-20s ┆ nbg1                        │
│ 9aee5840-246d-4a74-a60f-a6b6e0597ed7 ┆ infra-lmj8t-46c9c            ┆ ACTIVE ┆ SCS-2V-4-20s ┆ nbg3                        │
│ cc5c012d-4b66-43ac-9527-084f5458bdff ┆ infra-lmj8t-2hhpw            ┆ ACTIVE ┆ SCS-2V-4-20s ┆ nbg6                        │
└──────────────────────────────────────┴──────────────────────────────┴────────┴──────────────┴─────────────────────────────┘

[garloff]

Reviews would be nice ...

[Nils98Ar]

Is there a reason to use 3 separate variables instead of only one like volumesInSameAZ?

[garloff]

I change three different places and thus thought I need 3 different variables.

But you are right: I don't.
Whether or not the cloud uses AZ-specific storage is just one piece of information (and the tooling can detect it by looking at storage AZs and check whether it's just one or whether it's the same list as the compute AZs). And I apply it three times in conjunction with another individual variable, so it should be good to go with just one variable.
Will create an updated patch.

Thanks for the review, @Nils98Ar !

[garloff]

OK, so let's have only one flag, diskPinAZ to control all three places.
Reviews welcome!