fix(moodle): survive flock EIO on the NFS bootstrap lock and widen the converge budget

Six replicas race the bootstrap lock across three NFS clients; the truncating open invalidated the holder's lock state, so flock died with EIO and the task crash-looped with exit 65 until the winner finished the ~6min code copy. The lock file is now opened append-only and flock failures fall back to a sentinel recheck loop instead of killing PID 1. The converge budget gets the same 600-retry override the other slow-bootstrap roles already carry; the default 150 missed this run's recovery by six seconds.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

roles/web-app-moodle/files/build/moodle-entrypoint.sh

@@ -33,17 +33,15 @@ moodle_bootstrap_code_dir() {
   touch "${MOODLE_BOOTSTRAP_SENTINEL}"
 }
 
-if [ ! -f "${MOODLE_BOOTSTRAP_SENTINEL}" ]; then
-  mkdir -p "${MOODLE_CODE_DIR}"
-  exec 9>"${MOODLE_BOOTSTRAP_LOCK}"
-  flock 9
-  moodle_bootstrap_code_dir
-  flock -u 9
-  exec 9>&-
-fi
-
+mkdir -p "${MOODLE_CODE_DIR}"
 while [ ! -f "${MOODLE_BOOTSTRAP_SENTINEL}" ]; do
-  sleep 1
+  exec 9>>"${MOODLE_BOOTSTRAP_LOCK}"
+  if flock -w 30 9; then
+    moodle_bootstrap_code_dir
+  else
+    sleep 5
+  fi
+  exec 9>&-
 done
 
 if [ "$(id -u)" -eq 0 ] && [ "${1:-}" != "php-fpm" ]; then

roles/web-app-moodle/vars/main.yml

@@ -1,6 +1,7 @@
 ---
 application_id: "web-app-moodle"
 entity_name: "{{ application_id | get_entity_name }}"
+swarm_stack_converge_retries: 600
 
 MOODLE_CODE_DIR: "/var/www/html"
 MOODLE_DATA_DIR: "/var/www/moodledata"