fix: restart CSI controller and capture previous crash logs

ansible/roles/csi/tasks/main.yml

@@ -13,6 +13,15 @@
   command: kubectl apply -f {{ csi_manifest_url }}
   changed_when: true
 
+- name: Restart CSI controller to pick up current secret
+  command: kubectl -n kube-system rollout restart deployment/hcloud-csi-controller
+  changed_when: true
+
+- name: Wait for CSI controller deployment generation
+  command: kubectl -n kube-system rollout status deployment/hcloud-csi-controller --timeout=30s
+  failed_when: false
+  changed_when: false
+
 - name: Wait for CSI controller rollout
   command: kubectl rollout status deployment/hcloud-csi-controller -n kube-system --timeout={{ csi_rollout_timeout_seconds }}s
   register: csi_controller_rollout
@@ -43,6 +52,17 @@
   failed_when: false
   when: csi_controller_rollout.rc != 0
 
+- name: Describe CSI controller pod on failure
+  shell: |
+    pod="$(kubectl -n kube-system get pods -l app=hcloud-csi-controller -o jsonpath='{.items[0].metadata.name}' 2>/dev/null)"
+    if [ -n "$pod" ]; then
+      kubectl -n kube-system describe pod "$pod"
+    fi
+  register: csi_controller_pod_describe
+  changed_when: false
+  failed_when: false
+  when: csi_controller_rollout.rc != 0
+
 - name: Show CSI driver logs on failure
   command: kubectl -n kube-system logs deployment/hcloud-csi-controller -c hcloud-csi-driver --tail={{ csi_failure_log_tail_lines }}
   register: csi_driver_logs
@@ -70,9 +90,18 @@
       Deployment describe:
       {{ csi_controller_deploy_describe.stdout | default('n/a') }}
 
+      Pod describe:
+      {{ csi_controller_pod_describe.stdout | default('n/a') }}
+
       hcloud-csi-driver logs:
       {{ csi_driver_logs.stdout | default('n/a') }}
 
+      hcloud-csi-driver previous logs:
+      {{ csi_driver_previous_logs.stdout | default('n/a') }}
+
+      Sidecar previous logs:
+      {{ csi_sidecar_previous_logs.stdout | default('n/a') }}
+
       Recent kube-system events:
       {{ csi_recent_events.stdout | default('n/a') }}
   when: csi_controller_rollout.rc != 0
@@ -90,3 +119,27 @@
   fail:
     msg: "CSI node daemonset rollout failed: {{ csi_node_rollout.stdout | default('') }} {{ csi_node_rollout.stderr | default('') }}"
   when: csi_node_rollout.rc != 0
+- name: Show CSI driver previous logs on failure
+  shell: |
+    pod="$(kubectl -n kube-system get pods -l app=hcloud-csi-controller -o jsonpath='{.items[0].metadata.name}' 2>/dev/null)"
+    if [ -n "$pod" ]; then
+      kubectl -n kube-system logs "$pod" -c hcloud-csi-driver --previous --tail={{ csi_failure_log_tail_lines }}
+    fi
+  register: csi_driver_previous_logs
+  changed_when: false
+  failed_when: false
+  when: csi_controller_rollout.rc != 0
+
+- name: Show sidecar previous logs on failure
+  shell: |
+    pod="$(kubectl -n kube-system get pods -l app=hcloud-csi-controller -o jsonpath='{.items[0].metadata.name}' 2>/dev/null)"
+    if [ -n "$pod" ]; then
+      for container in csi-attacher csi-resizer csi-provisioner; do
+        echo "===== $container ====="
+        kubectl -n kube-system logs "$pod" -c "$container" --previous --tail={{ csi_failure_log_tail_lines }} || true
+      done
+    fi
+  register: csi_sidecar_previous_logs
+  changed_when: false
+  failed_when: false
+  when: csi_controller_rollout.rc != 0