Reorganized for new stable/waypoint versioning design.

cert-manager/versions/v1/scripts/repair-certificates.sh

@@ -0,0 +1,89 @@
+#!/bin/bash
+# Repair stuck certificates, orphaned ACME orders, and Cloudflare DNS errors.
+# This is an operational maintenance script, not part of deployment.
+# Run manually when cert-manager has issues with certificate issuance.
+#
+# Usage: KUBECONFIG=/path/to/kubeconfig ./repair-certificates.sh
+set -e
+set -o pipefail
+
+if [ -z "${KUBECONFIG}" ]; then
+    echo "ERROR: KUBECONFIG is not set"
+    exit 1
+fi
+
+needs_restart=false
+
+echo "=== cert-manager Certificate Repair ==="
+echo ""
+
+echo "Checking for certificates with failed issuance attempts..."
+stuck_certs=$(kubectl get certificates --all-namespaces -o json 2>/dev/null | \
+    jq -r '.items[] | select(.status.conditions[]? | select(.type=="Issuing" and .status=="False" and (.message | contains("404")))) | "\(.metadata.namespace) \(.metadata.name)"')
+
+if [ -n "$stuck_certs" ]; then
+    echo "WARNING: Found certificates stuck with non-existent orders, recreating them..."
+    echo "$stuck_certs" | while read ns name; do
+        echo "Recreating certificate $ns/$name..."
+        cert_spec=$(kubectl get certificate "$name" -n "$ns" -o json | jq '.spec')
+        kubectl delete certificate "$name" -n "$ns"
+        echo "{\"apiVersion\":\"cert-manager.io/v1\",\"kind\":\"Certificate\",\"metadata\":{\"name\":\"$name\",\"namespace\":\"$ns\"},\"spec\":$cert_spec}" | kubectl apply -f -
+    done
+    needs_restart=true
+    sleep 5
+else
+    echo "No certificates stuck with failed orders"
+fi
+
+echo "Checking for orphaned ACME orders..."
+orphaned_orders=$(kubectl logs -n cert-manager deployment/cert-manager --tail=200 2>/dev/null | \
+    grep -E "failed to retrieve the ACME order.*404" 2>/dev/null | \
+    sed -n 's/.*resource_name="\([^"]*\)".*/\1/p' | \
+    sort -u || true)
+
+if [ -n "$orphaned_orders" ]; then
+    echo "WARNING: Found orphaned ACME orders from logs"
+    for order in $orphaned_orders; do
+        echo "Deleting orphaned order: $order"
+        orders_found=$(kubectl get orders --all-namespaces 2>/dev/null | grep "$order" 2>/dev/null || true)
+        if [ -n "$orders_found" ]; then
+            echo "$orders_found" | while read ns name rest; do
+                kubectl delete order "$name" -n "$ns" 2>/dev/null || true
+            done
+        fi
+    done
+    needs_restart=true
+else
+    echo "No orphaned orders found in logs"
+fi
+
+echo "Checking for Cloudflare DNS cleanup errors..."
+cloudflare_errors=$(kubectl logs -n cert-manager deployment/cert-manager --tail=200 2>/dev/null | \
+    grep -c "Error: 7003.*Could not route" 2>/dev/null || echo "0")
+
+if [ "$cloudflare_errors" -gt "0" ]; then
+    echo "WARNING: Found $cloudflare_errors Cloudflare DNS cleanup errors (stale DNS record references)"
+    echo "Deleting stuck challenges and orders to allow fresh start"
+
+    kubectl delete challenges --all -n cert-manager 2>/dev/null || true
+    kubectl delete orders --all -n cert-manager 2>/dev/null || true
+
+    needs_restart=true
+else
+    echo "No Cloudflare DNS cleanup errors"
+fi
+
+if [ "$needs_restart" = true ]; then
+    echo "Restarting cert-manager to clear internal state..."
+    kubectl rollout restart deployment cert-manager -n cert-manager
+    kubectl rollout status deployment/cert-manager -n cert-manager --timeout=120s
+    echo "Waiting for cert-manager to recreate fresh challenges..."
+    sleep 15
+else
+    echo "No restart needed - cert-manager state is clean"
+fi
+
+echo ""
+echo "Repair complete. Check certificate status with:"
+echo "  kubectl get certificates --all-namespaces"
+echo "  kubectl get clusterissuers"