v0.6.0 — run history, job cancellation, Prometheus metrics

Added

Run history in the interface. The activity card now has two tabs. Alongside the familiar activity log there is a Run history table listing every run with its job type, what triggered it, the result, how long it took and how many guests it covered. Clicking a run expands it in place to show each step — wake, wait, backup, garbage collection, verify, power-off — with its own duration and outcome, plus that run's log lines. Failures show their error without expanding.

Stop a running job. The Run backup and Run GC buttons turn into a Stop button while a job is in progress, with a confirmation dialog that can also power the PBS off once the job has stopped. Stopping also cancels the underlying task on the Proxmox side, so a cancelled backup does not keep running on the server after Joulenap has stopped watching it. A scheduled verification can be stopped the same way. Previously a stuck backup or an unreachable PBS blocked every later job — and manual power-off — until the container was restarted.

Prometheus metrics for Grafana. A new /metrics endpoint, protected by the same read-only API key as the dashboard integration, exposing PBS power state, scheduler state, the last run's result and duration, datastore usage, run counts, and the last backup time of every individual guest. Scraping never wakes the PBS, and cached values keep reporting while it sleeps — so you can alert on a guest quietly dropping out of your backup set. docs/INTEGRATIONS.md has the scrape configuration, the full metric list, example queries and ready-made alert rules.

Changed

• Notifications name the job that ran. A scheduled verification or a garbage-collection run that failed used to notify "backup failed"; each job type now reports its own outcome, in English and Italian.
• Removed the backup.guests.auto_include_new setting. It never had any effect, while its name and default implied newly created guests were picked up automatically. Existing configuration files keep working — the key is ignored and dropped on the next save. The behaviour it seemed to promise is what "all" and "exclude" mode already do; "include" mode is, and always was, an explicit list.
• Documentation accuracy pass. Corrected the guest-selection and garbage-collection descriptions, the Proxmox VE token privilege list (which omitted Datastore.Audit and Datastore.Allocate, so a manually created token would fail at prune time), the Proxmox Backup Server token privileges, the supported-versions table, and the API reference, which was missing several endpoints. Added a walkthrough of the Settings tabs to the install guide.

Fixed

• Toggle switches are now announced correctly by screen readers, and can no longer submit a surrounding form by accident.
• Repaired three changelog comparison links that pointed at version tags which were never published.

Upgrading

No action needed. Existing config.yaml files load unchanged; the retired auto_include_new key is ignored and removed the next time settings are saved.

docker pull catubba/joulenap:0.6.0

Image digest: sha256:4f9657fb6c8b0b4698ae58931e56bb44416972bb03a6d205bccdd69600bd10b2

Full changelog: v0.5.0...v0.6.0