v1.0.513

Minor Release Released September 8, 2026

Notification system durability and bulk actions: delete-all/archive-all endpoints, persistent read/delete state.

Added a per-account disk warning threshold (resource_alert_threshold_disk), configurable next to CPU/RAM.
Notification content is now translatable in all 31 panel languages (title_key/message_key/i18n_params) instead of fixed English text.
Added "Delete all" and "Archive all" endpoints for notifications, with optional status/category filter.

Notification lifecycle (archive/delete) now matches the panel's stored preferences.
Removed dead settings with no backend effect: monitoring.alert_email/slack, backup email-notification fields, notification_sound, redis_cache_enabled.
Do-Not-Disturb is now evaluated in the panel's time zone; system notifications are delivered per recipient.

Feature-lock modal showed non-existent plan names ("Pro+"); plan names now come from the Central plan matrix.
Resource-alert flood fixed: per-account CPU is now measured against the account's plan limit instead of a raw multi-core host percentage; a single owner sends threshold alerts; kernel OOM/cgroup events are warning-only with a cooldown.
Alerts are now state-based: sent once when a threshold is crossed, again only on escalation, otherwise at most once per 24h, cleared when back under threshold. Fixes a disk-space warning repeating every 5 minutes.
Seeded event switches that were silently disabled: notify_root_login, notify_cpu_warning, notify_memory_warning, monitoring.resource_alerts_enabled. Event-switch map now covers warning+critical severities, SSL renewal failures, and resource alerts.
Notification grouping key now includes the title, so alerts about different accounts are never merged.
External API /v1/notifications now returns real data (ROOT-only) and mark-as-read works.
Notification mutations are now durable: deleted/read state no longer reverts after a backend restart.

All Releases