fix(backup): retry kubectl preflight when railiance01 API is flaky

Poll up to 10 times for forgejo and forgejo-db pods before failing
preflight when the k3s API returns ServiceUnavailable.

tools/cmd/forgejo-backup

@@ -26,8 +26,20 @@ FORGEJO_NAMESPACE="${FORGEJO_NAMESPACE:-forgejo}"
 FORGEJO_RELEASE="${FORGEJO_RELEASE:-forgejo}"
 FORGEJO_DB_NAMESPACE="${FORGEJO_DB_NAMESPACE:-databases}"
 FORGEJO_DB_CLUSTER="${FORGEJO_DB_CLUSTER:-forgejo-db}"
-FORGEJO_DB_POD="${FORGEJO_DB_POD:-$(kubectl get pods -n "${FORGEJO_DB_NAMESPACE}" \
+
-  -l "cnpg.io/cluster=${FORGEJO_DB_CLUSTER}" -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || true)}"
+kubectl_get_pod() {
+  local ns="$1" selector="$2" attempt out
+  for attempt in 1 2 3 4 5 6 7 8 9 10; do
+    out="$(kubectl get pods -n "${ns}" -l "${selector}" \
+      -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || true)"
+    if [[ -n "${out}" ]]; then
+      echo "${out}"
+      return 0
+    fi
+    sleep $((attempt * 3))
+  done
+  return 1
+}
 
 BACKUP_DIR="${XDG_CACHE_HOME:-$HOME/.cache}/railiance/backups/forgejo"
 TS="$(date -u +%Y%m%dT%H%M%SZ)"
@@ -46,6 +58,8 @@ trap cleanup_plain EXIT
 railiance_backup_require_tools
 railiance_backup_require_openbao_lane
 
+FORGEJO_DB_POD="${FORGEJO_DB_POD:-$(kubectl_get_pod "${FORGEJO_DB_NAMESPACE}" \
+  "cnpg.io/cluster=${FORGEJO_DB_CLUSTER}" || true)}"
 if [[ -z "${FORGEJO_DB_POD}" ]]; then
   bad "preflight" "forgejo-db pod not found (cluster ${FORGEJO_DB_CLUSTER})"
   exit 1
@@ -57,9 +71,8 @@ if ! kubectl wait --for=condition=Ready "pod/${FORGEJO_DB_POD}" \
   exit 1
 fi
 
-PROD_POD="$(kubectl get pods -n "${FORGEJO_NAMESPACE}" \
+PROD_POD="$(kubectl_get_pod "${FORGEJO_NAMESPACE}" \
-  -l "app.kubernetes.io/instance=${FORGEJO_RELEASE}" \
+  "app.kubernetes.io/instance=${FORGEJO_RELEASE}" || true)"
-  -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || true)"
 if [[ -z "${PROD_POD}" ]]; then
   bad "preflight" "Forgejo app pod not found in ${FORGEJO_NAMESPACE}"
   exit 1