On systems where Docker compatibility layer obscures Podman (like CoreOS),
the auto-detection can fail. Users can now force the runtime:
--docker-runtime podman
PULSE_DOCKER_RUNTIME=podman
Valid values: auto (default), docker, podman
Related to Discussion #958
Add PULSE_DISABLE_DOCKER_UPDATE_CHECKS environment variable and
--disable-docker-update-checks flag to disable Docker image update
detection. This is useful for:
- Avoiding Docker Hub rate limits
- Users who don't want update notifications in their dashboard
Related to Discussion #982
The /cluster/replication endpoint only returns job configuration (guest,
schedule, source, target), not status data (last_sync, next_sync,
duration, fail_count, state).
This fix enriches each replication job with status from the per-node
endpoint /nodes/{node}/replication/{id}/status to get timing and state
data needed for proper UI display.
Added integration tests to verify:
- Status endpoint is called and data is merged correctly
- Graceful handling when status endpoint fails
Fixes#992
On SELinux-enforcing systems (Fedora, RHEL, CentOS), binaries installed to
non-standard locations need proper security contexts for systemd to execute
them. Without this, systemd fails with 'Permission denied' even when the
binary has correct Unix permissions.
Changes:
- Add restore_selinux_contexts() function to both install scripts
- Uses restorecon (preferred) or chcon (fallback) to set bin_t context
- Only runs when SELinux is detected and enforcing
- Called after binary installation, before systemd service start
Previously, the PULSE_DISK_EXCLUDE environment variable and --disk-exclude
flag only filtered mount points in the hostmetrics collector. This change
extends the exclusion to SMART data collection.
Changes:
- Updated smartctl.CollectLocal() to accept diskExclude patterns
- Added matchesDeviceExclude() for block device pattern matching
- Patterns support: exact match (sda), prefix (nvme*), contains (*cache*)
- Updated hostagent to pass DiskExclude to SMART collector
- Added comprehensive tests for pattern matching
- Updated documentation
Previously, the TemperatureGauge component used hardcoded thresholds
(critical: 80°C, warning: 70°C) for text coloring. Now it uses the
user-configured temperature threshold from alert settings.
Changes:
- Add getTemperatureThreshold() helper to alertsActivation store
- Pass critical/warning props to TemperatureGauge in NodeSummaryTable
- Warning is set to (threshold - 5°C) matching the hysteresis pattern
When running as a unified agent (pulse-agent with --enable-docker), the
Docker module was using a different fallback chain for agent ID than the
host module. In unified mode with empty machineID, the Docker module fell
back to daemonID while the host module fell back to hostname. This caused
the server to reject Docker reports with 'token already in use by agent'
errors because the same API token was bound to different agent IDs.
The fix ensures that in unified mode, the Docker module uses the exact
same fallback chain as the host module: machineID -> hostname. The daemonID
fallback is only used in standalone mode for backward compatibility.
Fixes#985, #986
The runtime stage was copying both amd64 and arm64 pulse binaries to /tmp/,
then selecting one based on TARGETARCH and deleting the rest. Due to Docker's
immutable layers, the deleted binaries were still counted toward image size.
Changed to copy directly using TARGETARCH variable substitution, which only
copies the needed binary for the target architecture.
This saves ~34MB per architecture in the final image.
Note: The agent_runtime stage and /opt/pulse/bin/ download binaries still have
room for optimization, but require more complex changes.
Related to #981
When a powered-off VM is backed up by Proxmox, the alert briefly disappears
as the VM status changes. The previous fix (3830e701) preserved ackState when
alerts were removed, but the cleanup TTL was measured from the acknowledgement
time. For alerts acknowledged > 1 hour ago (common for intentionally powered-off
VMs), the ackState was immediately considered stale and deleted when cleanup ran.
The fix adds an inactiveAt timestamp to track when an alert was removed, and
uses this time for the cleanup TTL instead of the acknowledgement time. This
ensures acknowledgement state is preserved for at least 1 hour after the alert
disappears, regardless of when it was originally acknowledged.
Related to #980
- RAID tests now use /dev/md2 since md0/md1 are skipped for Synology compatibility
- AI handler tests now expect 'AI is not enabled' message after AI gating change
dispatchAlert() now checks if an alert is already acknowledged before sending
notifications. Previously, acknowledged alerts (especially backup-age alerts)
would continue to dispatch notifications every poll cycle because the
acknowledgement check was missing from the dispatch path.
The fix adds an early return in dispatchAlert() when alert.Acknowledged is true,
matching the existing checks for flapping, activation state, and quiet hours.
- Add IsAIEnabled() method to AISettingsHandler for consistent checks
- Gate baseline learning, pattern detector, and correlation detector initialization
in StartPatrol() on AI being enabled
- Add AI enabled checks to all /api/ai/intelligence/* endpoints as defense-in-depth
- Return empty results with "AI is not enabled" message when AI is disabled
This ensures no AI-related data is collected, persisted, or returned when AI is disabled,
preventing the "undismissable alerts" issue where old AI findings would appear.
The findMatchingHostAgent function now uses the proper linkedHostAgentId
relationship from the backend state, falling back to hostname matching only
if no linked agent is found. This fixes the 'Agent required' badge showing
incorrectly when agents are running and collecting data.
- Search for kubeconfig in /home/*/.kube/config in addition to /root/.kube/config
- Add --kubeconfig installer flag to specify custom kubeconfig path
- Auto-detect and pass kubeconfig path to agent when Kubernetes is enabled
- Respect KUBECONFIG environment variable when kubectl is working
Related to discussion #968
- Separated metrics collection into internal/dockeragent/collect.go
- Added agent self-update pre-flight check (--self-test)
- Implemented signed binary verification with key rotation for updates
- Added batch update support to frontend with parallel processing
- Cleaned up agent.go and added startup cleanup for backup containers
- Updated documentation for Docker features and agent security
When returning legacy arrays directly from store state, the createMemo
wouldn't notify dependents because the same array reference was returned
even when its contents changed (via reconcile). The spread operator
creates a new array reference each time, ensuring downstream memos
like sortedHosts() properly re-run.
This is a defense-in-depth fix complementing commit f5c0af77 which
fixed HostRow destructuring. Together these ensure reliable real-time
updates on the Hosts page.
Related to #949
- Add temperature unit preference in Settings > General
- Store preference in localStorage (temperatureUnit: 'celsius' | 'fahrenheit')
- Create formatTemperature() utility that respects the preference
- Update all temperature displays across the UI:
- TemperatureGauge component
- HostsOverview page (sensors, disk temps)
- DiskList (SMART temperatures)
- ThresholdSlider
- ThresholdBadge
- ThresholdsTable
- ProxmoxNodesSection
- NodeSummaryTable tooltips
- Alert formatters
When a host agent is running on a Proxmox node (linked host agent),
merge the agent's SMART disk temperature data into the Physical Disks
view for that node. This allows disk temps collected by pulse-agent
to populate the Physical Disks page without requiring Proxmox SMART
monitoring to be enabled.
Matching is done by WWN (most reliable), serial number, or device path.
Closes part of issue #909 (follow-up from MichiFr)
The SMART disk temperature data was being collected by the agent but not
passed through to the frontend. Fixed by:
1. Added SMART field to HostSensorSummaryFrontend and created
HostDiskSMARTFrontend type in models_frontend.go
2. Updated hostSensorSummaryToFrontend() in converters.go to include
SMART data conversion
3. Added HostDiskSMART interface to frontend TypeScript types
4. Updated HostTemperatureCell to display disk temperatures in tooltip
with a 'Disk Temperatures' section and fallback to SMART temps when
no CPU/sensor temps are available
Strip trailing slashes from PULSE_URL to prevent URLs like
http://host:7655//download/pulse-agent which incorrectly match
the frontend route instead of the download endpoint.
The sensor proxy self-heal script runs every 5 minutes and calls migrate-to-file.
Previously it would print 'Migration complete' every time, even when already in
file mode with nothing to migrate.
Now migrateInlineToFile returns a boolean indicating if migration actually
occurred, and the CLI only prints the message when work was done.
Synology NAS creates multiple shared folders (e.g., /volume1/docker, /volume1/photos)
that are all mount points on the same underlying BTRFS volume. Each reported the same
16TB total, causing Pulse to show 64TB+ instead of 16TB.
The fix tracks device+total combinations and only counts each unique pair once.
When duplicates are found, the shallowest mountpoint (e.g., /volume1) is preferred.
Added a unit test to verify the deduplication works correctly.
Previously, toggling AI Commands in the Agents view would show a pending state
and wait for the agent to confirm the change (up to 2 minutes). If the agent
was slow to report or the WebSocket update was missed, the toggle would appear
stuck.
Now, UpdateHostAgentConfig also updates the Host model in state immediately,
providing instant UI feedback. The agent will still receive the config on its
next report, but users see the change right away.
Added SetHostCommandsEnabled function to models.State for this purpose.
