Improve guestregister service diagnostics and recovery validation (#447)

Enhance rescue block by providing more detailed error reporting for cloudregister
and zypper logs, including new tracking for 'THROW' errors.

Update the final assertion to allow the task to succeed if the service restart
attempt successfully restores the registration state. This prevents false
positives for transient race conditions that were correctly handled during
the rescue phase.

Author: mpagot

ansible/playbooks/tasks/check-guestregister-service.yaml

@@ -44,10 +44,9 @@
       ansible.builtin.debug:
         msg:
           - "--- Rescue Block Entry Diagnosis for {{ inventory_hostname }} ---"
-          - "Task 'Wait for cloud guest registration' result for this host: {{ guestregister_service_state | default('VARIABLE NOT DEFINED') }}"
-          - "Individual host failure detected: {{ (guestregister_service_state.failed | default(false)) | ternary('YES', 'NO') }}"
-          - "Current any_errors_fatal setting: {{ any_errors_fatal | default('NOT SET (default is False)') }}"
-          - "Active hosts remaining in play: {{ ansible_play_hosts_all }}"
+          - "Task 'Wait for cloud guest registration' result for this host:"
+          - "           rc: {{ guestregister_service_state.rc | default('UNKNOWN') }}"
+          - "           stdout: {{ guestregister_service_state.stdout_lines | default('UNKNOWN') }}"
 
     # [DIAGNOSTIC] Initial service status and logs collection
     - name: Get guestregister.service status
@@ -72,19 +71,27 @@
     - name: Extract specific errors from cloudregister log
       ansible.builtin.shell: |
         set -o pipefail
-        grep -E "ERROR:|WARNING:|401 Unauthorized|422|Removing (repository artifacts|credentials|repo|service):" /var/log/cloudregister | tail -n 20
+        grep -E "ERROR:|WARNING:|401 (None|Unauthorized)|422|failed|(Deleting|Removing).*(repository artifacts|credentials|repo|service)|registration complete|No credentials entry" /var/log/cloudregister
       register: tailored_cloudregister_errors
       failed_when: false
       changed_when: false
 
     - name: Search for Zypper lock in zypper.log
       ansible.builtin.shell: |
         set -o pipefail
-        grep "System management is locked by the application with pid" /var/log/zypper.log | tail -n 5
+        grep "System management is locked by the application with pid" /var/log/zypper.log
       register: zypper_log_lock_error
       failed_when: false
       changed_when: false
 
+    - name: Search for Zypper THROW in zypper.log
+      ansible.builtin.shell: |
+        set -o pipefail
+        grep "THROW" /var/log/zypper.log
+      register: zypper_log_throw
+      failed_when: false
+      changed_when: false
+
     - name: Check repository state after failure
       ansible.builtin.command: zypper lr -u
       register: zypper_repos_after_fail
@@ -126,6 +133,7 @@
           - "--- Tailored Diagnostics for guestregister.service failure ---"
           - "Cloudregister Errors Summary: {{ tailored_cloudregister_errors.stdout_lines | default('None found') }}"
           - "Zypper Log Lock Errors: {{ zypper_log_lock_error.stdout_lines | default('None found') }}"
+          - "Zypper Log throws: {{ zypper_log_throw.stdout_lines | default('None found') }}"
           - "Racing PIDs and Commands: {{ racing_pids_discovery.stdout_lines | default('No racing PIDs identified') }}"
           - "Clean-up/Reset detected: {{ 'Removing' in (tailored_cloudregister_errors.stdout | default('')) }}"
           - "SMT entries in /etc/hosts: {{ hosts_smt_entries.stdout_lines | default('Not found') }}"
@@ -262,5 +270,8 @@
 
     - name: Fail if guestregister.service is not in the expected state
       ansible.builtin.assert:
-        that: false
+        that:
+          - "'ActiveState=inactive' in restart_state.stdout"
+          - "'Result=success' in restart_state.stdout"
+          - "not to_be_registered"
         fail_msg: "FATAL: guestregister.service is not in the expected state (inactive/success) after retries. Check logs above for details."