Our GCP worker pools (like this one) define a list of regions to use (currently just us-central1). We then define which zones get used in a region here:
https://hg.mozilla.org/ci/ci-configuration/file/0df1eb2b6e985b623152060139f4ebb701dfd021/environments.yml#l116

This appears to work, but I noticed that the SSDs are always set to the first zone in the list. For example, here's the config for the zone us-central1-b:

              {
                  "capacityPerInstance": 1,
                  "disks": [
                      {
                          "autoDelete": true,
                          "boot": true,
                          "initializeParams": {
                              "diskSizeGb": 20,
                              "sourceImage": "projects/taskcluster-imaging/global/images/fxci-level1-gcp-0y9sk1q1wfj5d06y0nea"
                          },
                          "type": "PERSISTENT"
                      },
                      {
                          "autoDelete": true,
                          "initializeParams": {
                              "diskType": "zones/us-central1-a/diskTypes/local-ssd"
                          },
                          "interface": "NVME",
                          "type": "SCRATCH"
                      },
                      {
                          "autoDelete": true,
                          "initializeParams": {
                              "diskType": "zones/us-central1-a/diskTypes/local-ssd"
                          },
                          "interface": "NVME",
                          "type": "SCRATCH"
                      },
                      {
                          "autoDelete": true,
                          "initializeParams": {
                              "diskType": "zones/us-central1-a/diskTypes/local-ssd"
                          },
                          "interface": "NVME",
                          "type": "SCRATCH"
                      },
                      {
                          "autoDelete": true,
                          "initializeParams": {
                              "diskType": "zones/us-central1-a/diskTypes/local-ssd"
                          },
                          "interface": "NVME",
                          "type": "SCRATCH"
                      }
                  ],
                  "machineType": "zones/us-central1-b/machineTypes/n2-custom-16-73728",
                  "minCpuPlatform": "Intel Cascadelake",
                  "networkInterfaces": [
                      {
                          "accessConfigs": [
                              {
                                  "type": "ONE_TO_ONE_NAT"
                              }
                          ]
                      }
                  ],
                  "region": "us-central1",
                  "scheduling": {
                      "automaticRestart": false,
                      "onHostMaintenance": "terminate",
                      "preemptible": true
                  },
                  "workerConfig": {
                      "capacity": 1,
                      "deviceManagement": {
                          "hostSharedMemory": {
                              "enabled": false
                          },
                          "kvm": {
                              "enabled": false
                          }
                      },
                      "shutdown": {
                          "afterIdleSeconds": 900,
                          "enabled": true
                      }
                  },
                  "zone": "us-central1-b"
              },

Notice the diskType key has us-central1-a listed under it. This seems to be happening because we're hardcoding the us-central1-a into the SSD configs directly in worker-pools.yml.

I'm not sure if this is a problem or not, but I'm assuming (without looking it up) that we'll want to use disks from the same zone we're running tasks in.

Should be all fixed now per the most recent FxCI worker pool config.