refactor: use gcloud to search for cached disk state (#3775)

.github/workflows/test.yml

@@ -194,7 +194,6 @@ jobs:
     runs-on: ubuntu-latest
     needs: build
     outputs:
-      disk_short_sha: ${{ steps.disk-short-sha.outputs.disk_short_sha }}
       any_changed: ${{ steps.changed-files-specific.outputs.any_changed }}
     steps:
       - uses: actions/checkout@v2.4.0
@@ -306,22 +305,6 @@ jobs:
           --storage-location=us \
           --description="Created from head branch ${{ env.GITHUB_HEAD_REF_SLUG_URL }} targeting ${{ env.GITHUB_BASE_REF_SLUG }} from PR ${{ env.GITHUB_REF_SLUG_URL }} with commit ${{ env.GITHUB_EVENT_PULL_REQUEST_HEAD_SHA }}"
 
-      - name: Output and write the disk SHORT_SHA to a txt
-        if: steps.sync-to-checkpoint.outcome == 'success'
-        id: disk-short-sha
-        run: |
-          short_sha=$(echo "${{ env.GITHUB_SHA_SHORT }}")
-          echo "$short_sha" > latest-disk-state-sha.txt
-          echo "::set-output name=disk_short_sha::$short_sha"
-
-      - name: Upload the disk state txt
-        if: steps.sync-to-checkpoint.outcome == 'success'
-        uses: actions/upload-artifact@v2.3.1
-        with:
-          name: latest-disk-state-sha
-          path: latest-disk-state-sha.txt
-          retention-days: 1095
-
       - name: Delete test instance
         # Do not delete the instance if the sync timeouts in GitHub
         if: ${{ steps.sync-to-checkpoint.outcome == 'success' || steps.sync-to-checkpoint.outcome == 'failure' }}
@@ -346,26 +329,6 @@ jobs:
         run: |
           echo LOWER_NET_NAME="${{ github.event.inputs.network || env.NETWORK }}" | awk '{print tolower($0)}' >> $GITHUB_ENV
 
-      # Get the latest uploaded txt with the disk SHORT_SHA from this workflow
-      - name: Download latest disk state SHORT_SHA
-        uses: dawidd6/action-download-artifact@v2.17.0
-        # Just search for the latest uploaded artifact if the previous disk regeneration job was skipped,
-        # otherwise use the output from ${{ needs.regenerate-stateful-disks.outputs.disk_short_sha }}
-        if: ${{ needs.regenerate-stateful-disks.outputs.any_changed != 'true' || github.event.inputs.regenerate-disks != 'true'}}
-        with:
-          github_token: ${{ secrets.GITHUB_TOKEN }}
-          workflow: test.yml
-          workflow_conclusion: ''
-          name: latest-disk-state-sha
-          check_artifacts:  true
-
-      - name: Get disk state SHA from txt
-        id: get-disk-sha
-        if: ${{ needs.regenerate-stateful-disks.outputs.any_changed != 'true' || github.event.inputs.regenerate-disks != 'true'}}
-        run: |
-          output=$(cat latest-disk-state-sha.txt)
-          echo "::set-output name=sha::$output"
-
       # Setup gcloud CLI
       - name: Authenticate to Google Cloud
         id: auth
@@ -373,6 +336,15 @@ jobs:
         with:
           credentials_json: ${{ secrets.GOOGLE_CREDENTIALS }}
 
+      - name: Get disk state name from gcloud
+        id: get-disk-name
+        if: ${{ needs.regenerate-stateful-disks.outputs.any_changed != 'true' || github.event.inputs.regenerate-disks != 'true'}}
+        run: |
+          output=$(gcloud compute images list --filter="zebrad-cache" --format="value(NAME)" --sort-by=~creationTimestamp --limit=1)
+          echo "Disk: $output"
+          echo "Description: $(gcloud compute images describe $output --format='value(DESCRIPTION)')"
+          echo "::set-output name=sha::$output"
+
       # Creates Compute Engine virtual machine instance w/ disks
       - name: Create GCP compute instance
         id: create-instance
@@ -380,7 +352,7 @@ jobs:
           gcloud compute instances create-with-container "sync-checkpoint-${{ env.GITHUB_REF_SLUG_URL }}-${{ env.GITHUB_SHA_SHORT }}" \
           --boot-disk-size 100GB \
           --boot-disk-type pd-ssd \
-          --create-disk=image=zebrad-cache-${{ env.DISK_SHORT_SHA }}-${{ env.lower_net_name }}-canopy,name=zebrad-cache-${{ env.GITHUB_SHA_SHORT }}-${{ env.lower_net_name }}-canopy,size=100GB,type=pd-ssd \
+          --create-disk=image=${{ env.CACHED_DISK_NAME }},name=zebrad-cache-${{ env.GITHUB_SHA_SHORT }}-${{ env.lower_net_name }}-canopy,size=100GB,type=pd-ssd \
           --container-mount-disk=mount-path='/zebrad-cache',name=zebrad-cache-${{ env.GITHUB_SHA_SHORT }}-${{ env.lower_net_name }}-canopy \
           --container-image ${{ env.GAR_BASE }}/${{ env.IMAGE_NAME }}:sha-${{ env.GITHUB_SHA_SHORT }} \
           --container-restart-policy=never \
@@ -402,7 +374,7 @@ jobs:
           --tags zebrad \
           --zone "${{ env.ZONE }}"
         env:
-          DISK_SHORT_SHA: ${{ needs.regenerate-stateful-disks.outputs.disk_short_sha || steps.get-disk-sha.outputs.sha }}
+          CACHED_DISK_NAME: ${{ steps.get-disk-name.outputs.sha }}
 
       # TODO: this approach is very mesy, but getting the just created container name is very error prone and GCP doesn't have a workaround for this without requiring a TTY
       # This TODO relates to the following issues: