Improve dashboard health and state recovery

This commit is contained in:
Mikei386
2026-07-13 21:06:09 +02:00
parent 819055cbf0
commit 8f75a04db6
15 changed files with 313 additions and 38 deletions
+7 -1
View File
@@ -2,13 +2,19 @@
<!DOCTYPE PLUGIN [
<!ENTITY name "urbm">
<!ENTITY author "Michael Roll">
<!ENTITY version "2026.07.13.r011">
<!ENTITY version "2026.07.13.r012">
<!ENTITY pluginURL "https://git.casaderoll.de/michael/URBM/raw/branch/main/dist/urbm.plg">
<!ENTITY packageURL "https://git.casaderoll.de/michael/URBM/raw/branch/main/dist/urbm-&version;-x86_64-1.txz">
<!ENTITY packageSHA256 "REPLACE_DURING_RELEASE">
]>
<PLUGIN name="&name;" author="&author;" version="&version;" pluginURL="&pluginURL;" min="7.0.0" support="https://git.casaderoll.de/michael/URBM/issues" icon="urbm.png">
<CHANGES>
### 2026.07.13.r012
- Draw a separate color-coded line and legend entry for every backup job when the dashboard filter is set to all jobs.
- Base dashboard health on successful backups, failed backup attempts, active work, and timezone-aware overdue schedules instead of unrelated maintenance or restore successes.
- Keep the WebGUI available when metrics, activities, health, or dashboard-status requests fail independently, with localized errors and retry actions.
- Quarantine malformed run and metric state files with a timestamped `.corrupt-*` suffix before creating clean replacements, and log the recovery path.
### 2026.07.13.r011
- Use durable backup measurements when deciding whether a scheduled job needs a catch-up run, even after activity history was cleared.
- Append one compact JSON-lines record per completed backup instead of rewriting the full metric history for every queue state change.