chore: enhance Supabase retained disk cleanup with project resolution and detailed skip reasons

- Updated `deploy.sh` to resolve project IDs from multiple configuration fallbacks and environment contexts for retained disk cleanup.
- Added support for `supabase_cfg_first_nonempty_value` to prioritize configuration over defaults.
- Enhanced logging with explicit reasons for skipped disk cleanup actions (e.g., missing gcloud, unresolved project).
- Updated tests to validate fallback logic and skip reason reporting.

scripts/cleanup-supabase-pvs.sh

@@ -1,44 +1,109 @@
 #!/usr/bin/env bash
 set -euo pipefail
 
-PROJECT="plenary-truck-485623-p7"
+PROJECT="${PROJECT:-${GCP_PROJECT_ID:-${GOOGLE_CLOUD_PROJECT:-plenary-truck-485623-p7}}}"
+DRY_RUN_RAW="${SUPABASE_RETAINED_PV_CLEANUP_DRY_RUN:-false}"
+DRY_RUN="false"
+if [[ "${DRY_RUN_RAW,,}" =~ ^(1|true|yes|on)$ ]]; then
+  DRY_RUN="true"
+fi
 
-echo "==> Finding stale Supabase PVs..."
+echo "==> Finding stale Supabase PVs (Released + Retain, Supabase namespace/storageClass)..."
+echo "==> Dry run: ${DRY_RUN}"
+echo "==> Project: ${PROJECT}"
 
-kubectl get pv -o json 
-| jq -r '
-  .items[]
-  | select(
-      .status.phase=="Released"
-      and .spec.storageClassName=="supabase-standard"
-    )
-  | [
-      .metadata.name,
-      .spec.csi.volumeHandle
-    ]
-  | @tsv
-' | while IFS=$'\t' read -r PV DISK; do
+mapfile -t CANDIDATES < <(
+  kubectl get pv -o json | jq -r '
+    .items[]
+    | . as $pv
+    | ($pv.spec.claimRef.namespace // "") as $claimNs
+    | ($pv.spec.storageClassName // "") as $sc
+    | select(
+        ($pv.status.phase == "Released")
+        and (($pv.spec.persistentVolumeReclaimPolicy // "") == "Retain")
+        and (
+          $sc == "supabase-standard"
+          or $claimNs == "supabase"
+        )
+      )
+    | (
+        ($pv.spec.csi.volumeHandle // $pv.spec.gcePersistentDisk.pdName // "")
+      ) as $handle
+    | (
+        ($pv.metadata.labels["topology.kubernetes.io/zone"]
+          // $pv.metadata.labels["failure-domain.beta.kubernetes.io/zone"]
+          // $pv.spec.csi.volumeAttributes["topology.gke.io/zone"]
+          // (
+            if ($handle | test("/zones/[^/]+/disks/[^/]+$"))
+            then ($handle | capture("/zones/(?<z>[^/]+)/disks/").z)
+            else ""
+            end
+          )
+        )
+      ) as $zone
+    | [
+        ($pv.metadata.name // ""),
+        $claimNs,
+        ($pv.spec.claimRef.name // ""),
+        ($pv.status.phase // ""),
+        ($pv.spec.persistentVolumeReclaimPolicy // ""),
+        $sc,
+        $handle,
+        $zone
+      ]
+    | @tsv
+  '
+)
 
-  if [[ -z "$DISK" || "$DISK" == "null" ]]; then
-    echo "[WARN] PV $PV has no disk handle, skipping"
+if [[ ${#CANDIDATES[@]} -eq 0 ]]; then
+  echo "==> No stale Supabase Released/Retain PV candidates found."
+  exit 0
+fi
+
+echo "==> Candidates:"
+for row in "${CANDIDATES[@]}"; do
+  IFS=$'\t' read -r PV CLAIM_NS CLAIM_NAME PHASE RECLAIM SC DISK ZONE <<<"$row"
+  echo "  - PV=${PV} claim=${CLAIM_NS}/${CLAIM_NAME:-<none>} phase=${PHASE} reclaim=${RECLAIM} sc=${SC} disk=${DISK:-<none>} zone=${ZONE:-<unknown>}"
+done
+
+for row in "${CANDIDATES[@]}"; do
+  IFS=$'\t' read -r PV CLAIM_NS CLAIM_NAME PHASE RECLAIM SC DISK ZONE <<<"$row"
+  NAME="${DISK##*/}"
+
+  if [[ "$DRY_RUN" == "true" ]]; then
+    echo "[dry-run] kubectl delete pv ${PV} --ignore-not-found"
+    if [[ -n "$DISK" && "$DISK" != "null" ]]; then
+      if [[ -n "$ZONE" ]]; then
+        echo "[dry-run] gcloud compute disks delete ${NAME} --zone ${ZONE} --project ${PROJECT} --quiet"
+      else
+        echo "[dry-run] skip disk delete for ${DISK} (missing zone)"
+      fi
+    else
+      echo "[dry-run] skip disk delete for PV ${PV} (no disk handle)"
+    fi
     continue
   fi
 
-  # extract zone + disk name
-  ZONE=$(echo "$DISK" | awk -F/ '{print $(NF-2)}')
-  NAME=$(echo "$DISK" | awk -F/ '{print $NF}')
-
-  echo "==> Cleaning PV=$PV disk=$NAME zone=$ZONE"
-
-  echo "  -> deleting PV"
+  echo "==> Deleting PV ${PV}"
   kubectl delete pv "$PV" --ignore-not-found
 
-  echo "  -> deleting GCE disk"
-  gcloud compute disks delete "$NAME" 
-    --zone "$ZONE" 
-    --project "$PROJECT" 
-    --quiet || echo "[WARN] disk delete failed: $NAME"
+  if [[ -z "$DISK" || "$DISK" == "null" ]]; then
+    echo "[WARN] PV ${PV} has no disk handle; skipped disk delete"
+    continue
+  fi
+  if [[ -z "$ZONE" ]]; then
+    echo "[WARN] Disk ${DISK} zone is unknown; skipped disk delete"
+    continue
+  fi
 
+  if ! command -v gcloud >/dev/null 2>&1; then
+    echo "[WARN] gcloud not found; skipped disk delete for ${DISK}"
+    continue
+  fi
+
+  echo "==> Deleting GCE disk ${NAME} (zone=${ZONE})"
+  gcloud compute disks delete "$NAME" --zone "$ZONE" --project "$PROJECT" --quiet \
+    || echo "[WARN] disk delete failed: ${NAME}"
 done
 
 echo "==> Cleanup complete"