update 1.2.4.1

AppImage/scripts/flask_server.py

@@ -10048,7 +10048,12 @@ def _update_smart_cron():
         # scheduled tests in history" with no obvious clue why.
         cmd = f'/usr/local/share/proxmenux/scripts/storage/smart-scheduled-test.sh --schedule-id {schedule_id} --test-type {test_type} --retention {retention}'
         if disks != ['all']:
-            cmd += f" --disks '{','.join(disks)}'"
+            # Prepend /dev/ when the UI stored the disk as a basename
+            # (sdb, nvme0n1, …). The runner's `[[ -b "$disk" ]]` check
+            # only accepts full block-device paths, so basenames were
+            # silently skipped and no test ever ran.
+            disk_args = [d if d.startswith('/') else f'/dev/{d}' for d in disks]
+            cmd += f" --disks '{','.join(disk_args)}'"
         
         cron_lines.append(f'{cron_time} root {cmd} >> /var/log/proxmenux/smart-schedule.log 2>&1')
     

scripts/backup_restore/run_scheduled_backup.sh

@@ -547,19 +547,54 @@ main() {
   # jobs keep whatever KEEP_* the .env has.
   _sb_hydrate_attached_retention
 
-  local lock_file="${LOCK_DIR}/proxmenux-backup-${job_id}.lock"
-  if command -v flock >/dev/null 2>&1; then
-    exec 9>"$lock_file" || exit 1
-    if ! flock -n 9; then
-      echo "Another run is active for job ${job_id}" >&2
-      exit 1
-    fi
-  fi
-
+  # Create log_file and summary_file BEFORE the flock so any early
+  # failure (lock contention, missing tools, unreadable env) surfaces
+  # in the runner log the Monitor polls, instead of being lost to
+  # stderr/DEVNULL and leaving the UI stuck on "Waiting for runner
+  # to start…" forever.
   local ts log_file stage_root summary_file
   ts="$(date +%Y%m%d_%H%M%S)"
   log_file="${LOG_DIR}/${job_id}-${ts}.log"
   summary_file="${LOG_DIR}/${job_id}-last.status"
+
+  local lock_file="${LOCK_DIR}/proxmenux-backup-${job_id}.lock"
+  if command -v flock >/dev/null 2>&1; then
+    if ! exec 9>"$lock_file"; then
+      {
+        echo "=== Scheduled backup job ${job_id} aborted at $(date -Iseconds) ==="
+        echo "Cannot open lock file: $lock_file"
+        echo "Check that ${LOCK_DIR} exists and is writable by root."
+      } >"$log_file"
+      {
+        echo "JOB_ID=${job_id}"
+        echo "RUN_AT=$(date -Iseconds)"
+        echo "RESULT=failed"
+        echo "REASON=lock_open_failed"
+      } >"$summary_file"
+      exit 1
+    fi
+    if ! flock -n 9; then
+      {
+        echo "=== Scheduled backup job ${job_id} aborted at $(date -Iseconds) ==="
+        echo "Another run is already active for this job (lock held: $lock_file)."
+        echo ""
+        echo "Possible causes:"
+        echo "  - A previous run is still in progress (check with: ps auxf | grep run_scheduled_backup)."
+        echo "  - A previous run hung and left the lock behind. Kill the stale"
+        echo "    process (if any) and remove the lock: rm $lock_file"
+        echo ""
+        echo "This run did NOT execute a backup."
+      } >"$log_file"
+      {
+        echo "JOB_ID=${job_id}"
+        echo "RUN_AT=$(date -Iseconds)"
+        echo "RESULT=failed"
+        echo "REASON=another_run_active"
+      } >"$summary_file"
+      exit 1
+    fi
+  fi
+
   stage_root="$(mktemp -d /tmp/proxmenux-sched-stage.XXXXXX)"
 
   {

scripts/storage/smart-scheduled-test.sh

@@ -169,6 +169,13 @@ FAIL_COUNT=0
 log "Found $TOTAL_DISKS disk(s) to test"
 
 for disk in $DISK_LIST; do
+    # Accept both full paths and basenames. Older cron entries written
+    # before the Monitor prepended /dev/ passed disk names as bare
+    # basenames (sdb, nvme0n1), which then failed the block-device
+    # check below. Normalising here keeps those legacy entries working
+    # after the update without forcing the operator to recreate them.
+    [[ "$disk" != /* ]] && disk="/dev/$disk"
+
     # Skip if disk doesn't exist
     if [[ ! -b "$disk" ]]; then
         log "WARNING: Disk $disk not found, skipping"