- Add PBS alert monitoring (CPU, memory, offline detection)
- Add storage offline detection with proper cluster awareness
- Remove bulk toggle feature from thresholds UI (unnecessary complexity)
- Add enable/disable buttons for PBS servers in thresholds tab
- Fix storage offline detection to avoid false positives in clusters
(only alert on truly offline storage, not inactive cluster storage)
Alert improvements:
- PBS instances now properly monitored like nodes
- Storage devices generate offline alerts with confirmation system
- All resource types support custom thresholds and disable toggles
- Consistent alert ID format across all resource types
- Proper hysteresis and confirmation counts to prevent flapping
addresses #123 (if there was an issue about missing PBS alerts)
Storage devices with alerts disabled in the Thresholds tab were still
triggering alerts. Added proper checking of the disabled override flag
in CheckStorage() to match the behavior of guest alerts.
- Changed default thresholds for diskRead/diskWrite/networkIn/networkOut to 0 (Off)
- UI now shows "Off" instead of "0 MB/s" for disabled I/O metrics
- Reduces alert noise for users who don't need I/O monitoring
- Advanced users can still enable them by setting custom thresholds
- Prevents common false positives like hourly pihole database writes
This makes Pulse focus on the core metrics (CPU, memory, disk space) by default
while keeping I/O monitoring available for those who need it.
- Fixed alert units display (MB/s vs %) in notifications and UI
- Added missing threshold controls for disk I/O and network metrics
- Redesigned threshold table with active alert indicators
- Added ability to disable node connectivity alerts
- Improved visual distinction between defaults and overrides
- Added keyboard shortcuts for search (type to search, ESC to clear)
- Better threshold management with automatic cleanup of empty overrides
- Enhanced UI with proper units display for all metrics
addresses #336
- Add ability to completely disable alerts for specific guests in Custom Overrides
- Refactor override editing to use single form instead of inline editing
- Add dashboard indicators for guests with custom overrides (blue cog for custom thresholds, grey bell-slash for disabled)
- Remove complex Proxmox tag-based alert control system in favor of simpler UI controls
- Improve layout and UX for alert override management
removed incorrect 'temporary override' framing - users want permanent control for VMs with special needs (TrueNAS, Samba, Frigate, etc). tags are now presented as direct per-VM control that can be used for both permanent configuration and temporary needs
key insight: tags let users say 'this specific VM is special' vs custom rules for patterns
- updated documentation to clearly distinguish between:
- Custom Rules: permanent alert policies configured in UI
- Tags: temporary operational overrides applied in Proxmox
- improved log messages to indicate tags are temporary overrides
- added clear use cases and best practices for each approach
- emphasized that tags complement (not replace) custom rules
changed pulse-relaxed behavior to override with fixed values (95% CPU/RAM, 98% disk) rather than adding to existing thresholds. this avoids confusing interactions with custom alert rules and provides more predictable behavior
also updated docs to clarify the priority order of tags vs custom rules
addresses #336 - diskRead/diskWrite/networkIn/networkOut metrics are measured in MB/s not percentages, so alert messages now correctly show "X MB/s" instead of "X%"
- pulse-no-alerts: suppress all alerts for VM/CT
- pulse-monitor-only: show alerts in UI but skip email notifications
- pulse-relaxed: use relaxed thresholds (+15% for CPU/memory, +10% for disk)
Tags are read from Proxmox VM/CT configuration and applied during alert checks.
This provides a simple way to manage alert behavior without adding UI complexity.
Requires 3 consecutive offline polls (~15 seconds) before triggering connectivity alerts to prevent false positives from transient cluster communication issues
- Fixed double CPU percentage multiplication for containers/VMs
- Added CheckGuest calls to efficient polling path
- Fixed newline escaping in grouped webhook notifications
- Guest alerts now properly trigger for containers and VMs
These changes address issues where guest alerts weren't being triggered
at all due to the efficient polling path not calling CheckGuest, and
webhook notifications were failing due to unescaped newlines in grouped
alert messages breaking JSON templates.
- Fixed duplicate 'usage' text in storage alerts (Storage usage usage → Storage at X%)
- Improved grouped alerts to show details of other alerts instead of just count
- Added severity indicators (🔴 critical, 🟡 warning, 🟢 info)
- Better formatting with current value, threshold, type, duration, and timestamp
- Cleaner message structure with proper Markdown formatting
- Shows up to 3 additional alerts in grouped notifications with their values
- alerts now trigger when nodes go offline or lose connectivity
- alerts automatically resolve when nodes come back online
- offline status is checked before resource thresholds
- logs critical events when nodes lose connectivity
This was a major oversight - nodes going offline should always trigger alerts
- Fixed Docker entrypoint to properly handle running as root (PUID=0)
- Improved alert history loading to handle permission errors gracefully
- Container now correctly runs as root when PUID=0 is set
- Alert history continues loading even if backup file has permission issues
Addresses #266 and #262
- Fix CPU core display to show for all guests with CPU data
- Previously only showed cores when CPU > 0 (truthy)
- Now shows "(0.0/X cores)" consistently for all running/stopped guests
- Improve code organization with new helper utilities
- Clean up import statements and remove debug logs
- Removed PBS summary card from Dashboard and Backups tabs (not needed)
- Fixed backup frequency chart to use local timezone instead of UTC
- Chart now properly includes today in the date range
- Dates display according to user's browser timezone
The issue was that the binary was looking for frontend files relative to the current working directory instead of relative to the binary location. This caused a redirect loop when the binary was installed in /opt/pulse but run from elsewhere.
Fixed by:
- Making the binary detect its own location and look for frontend files there
- Adding fallback paths for common installation locations
- Logging where frontend files are being served from for debugging
- Add GetDataDir() function to respect PULSE_DATA_DIR environment variable
- Update all hardcoded /var/lib/pulse paths to use configurable data directory
- Fix circular import by moving GetDataDir to utils package
- Ensures Docker containers can properly persist configuration and alerts
- Active alerts now persist to /var/lib/pulse/alerts/active-alerts.json
- Alerts are saved when created, resolved, and every minute
- On startup, alerts are restored with original timestamps
- Prevents duplicate email notifications after service restarts
- Skips restoration of alerts older than 24 hours
- Preserves acknowledgment state and escalation levels
- Remove recently resolved alerts from frontend display
- Keep recently resolved tracking in backend for potential future use
- Update alert indicators to show only critical and warning states
- Remove fade animations and resolved alert styling
- Improve dashboard clarity by focusing only on current issues
The dashboard now provides a cleaner, more focused view showing only
alerts that require immediate attention.
- Fixed alert clearing logic to work even when alerts are acknowledged
- Added immediate WebSocket state broadcast after alert resolution
- Fixed frontend activeAlerts store updates to maintain SolidJS reactivity
- Added logging for alert resolution events
The alert system now properly:
- Creates alerts when thresholds are exceeded
- Clears alerts automatically when values drop below clear threshold
- Updates frontend in real-time without requiring page refresh
- Go backend with Proxmox/PBS integration
- Modern TypeScript/SolidJS frontend
- WebSocket real-time updates
- Clean project structure with no legacy code
Pulse Monitor