- Add FeatureLongTermMetrics license feature for Pro tier
- Implement tiered storage in metrics store (raw, minute, hourly, daily)
- Add covering index for unified history query performance
- Seed mock data for 90 days with appropriate aggregation tiers
- Update PULSE_PRO.md to document the feature
- 7-day history remains free, 30d/90d requires Pro license
Remove proxy-related temperature code paths:
- temperature.go: remove proxy client integration and fallback logic
- config.go: remove SensorProxyEnabled and related config fields
- monitor.go: remove proxy client initialization and state
Temperature monitoring now relies solely on the unified agent approach.
The sensor proxy approach for temperature monitoring has been superseded
by the unified agent architecture where host agents report temperature
data directly. This removes:
- cmd/pulse-sensor-proxy/ - standalone proxy daemon
- internal/tempproxy/ - client library
- internal/api/*temperature_proxy* - API handlers and tests
- internal/api/sensor_proxy_gate* - feature gate
- internal/monitoring/*proxy_test* - proxy-specific tests
- scripts/*sensor-proxy* - installation and management scripts
- security/apparmor/, security/seccomp/ - proxy security profiles
Temperature monitoring remains available via the unified agent approach.
- Add PendingUpdates and PendingUpdatesCheckedAt fields to Node model
- Add GetNodePendingUpdates method to Proxmox client (calls /nodes/{node}/apt/update)
- Add 30-minute polling cache to avoid excessive API calls
- Add pendingUpdates to frontend Node type
- Add color-coded badge in NodeSummaryTable (yellow: 1-9, orange: 10+)
- Update test stubs for interface compliance
Requires Sys.Audit permission on Proxmox API token to read apt updates.
GetHostAgentConfig was loading profiles and assignments from disk on
every agent report (every 10-30 seconds per host). With multiple hosts,
this caused disk I/O contention that eventually led to request timeouts.
Added in-memory caching with 60-second TTL:
- Fast path reads from cache without locks when valid
- Double-checked locking pattern for cache refresh
- Cache auto-invalidates after TTL, no manual invalidation needed
Adds automatic Docker detection for Proxmox LXC containers:
- New HasDocker and DockerCheckedAt fields on Container model
- Docker socket check via connected agents on first run, restart, or start
- Parallel checking with timeouts for efficiency
- Caches results and only re-checks after state transitions
This enables the AI to know which LXC containers are Docker hosts
for better infrastructure guidance.
The agent was crashing with 'fatal error: concurrent map writes' when
handleCheckUpdatesCommand spawned a goroutine that called collectOnce
concurrently with the main collection loop. Both code paths access
a.prevContainerCPU without synchronization.
Added a.cpuMu mutex to protect all accesses to prevContainerCPU in:
- pruneStaleCPUSamples()
- collectContainer() delete operation
- calculateContainerCPUPercent()
Related to #1063
Users with removable/unmounted datastores (e.g., external HDDs for
offline backup) experienced excessive PBS log entries because Pulse
was querying all datastores including unavailable ones.
Added `excludeDatastores` field to PBS node configuration that accepts
patterns to exclude specific datastores from monitoring:
- Exact names: "exthdd1500gb"
- Prefix patterns: "ext*"
- Suffix patterns: "*hdd"
- Contains patterns: "*removable*"
Pattern matching is case-insensitive.
Fixes#1105
The previous fix (4090d981) addressed memory reporting for Docker agents
running in LXC containers, but the same issue also affects host agents.
When gopsutil runs inside an LXC, it can read Total and Free memory
from cgroup limits, but reports 0 for Used memory.
Added the same Total - Free fallback calculation to the host agent
processing path, which populates the Hosts tab.
Fixes#1075
When running Docker inside an LXC container, gopsutil can read the
Total memory (from cgroup limits) and Free memory correctly, but
returns 0 for Used memory. This caused the display to show "0B / 7GB"
even though memory was being used.
Added a fallback that calculates Used = Total - Free when Used is 0
but Total and Free are valid. This completes the fallback chain for
Docker-in-LXC memory reporting.
Fixes#1075
Previously, when some cluster endpoints were unreachable (e.g., backup
nodes intentionally offline), the cluster was marked as "degraded" even
though the Proxmox cluster itself was healthy and had quorum.
Now the connection health check queries the Proxmox cluster's actual
quorum status. A cluster is only marked "degraded" if it has lost
quorum (not enough votes for consensus), which is the actual indicator
of cluster instability.
This means:
- Cluster with quorum + some nodes offline = "healthy"
- Cluster without quorum = "degraded" (warning)
- All endpoints down = "error"
Fixes#1085
When multiple PVE nodes have the same hostname (e.g., both named "pve"),
auto-linking would incorrectly link all host agents to the first matching
node, causing temperature and sensor data to be mixed/duplicated.
Changes:
- findLinkedProxmoxEntity now detects hostname collisions and refuses
to auto-link, logging a warning instead
- Added manual link API endpoint (POST /api/agents/host/link) so users
can explicitly link agents to the correct nodes
- Added State.LinkHostAgentToNode for bidirectional manual linking
Fixes#1081
The previous fix (4ff9e58c) added a fallback for TotalMemoryBytes in
the agent when Docker's info.MemTotal returns 0 in LXC environments.
However, the server was not using TotalMemoryBytes to populate the
memory.Total field - it only used gopsutil's Memory.TotalBytes.
When gopsutil also fails to read memory in the LXC container, the
frontend would see memory.Total=0 and wouldn't fall back to
totalMemoryBytes due to JavaScript's nullish coalescing (??) only
triggering on null/undefined, not on 0.
This fix ensures the server uses TotalMemoryBytes as a fallback for
memory.Total when gopsutil returns 0, providing a complete fix chain:
1. Agent: Falls back to gopsutil when Docker returns 0
2. Server: Falls back to TotalMemoryBytes when gopsutil returns 0
Fixes#1075
Previously, when memory ballooning was active on a VM, Pulse would use
the balloon value as the total memory instead of the configured MaxMem.
This caused confusing displays where a 4GB VM with 1GB balloon would
show "94% (966MB/1GB)" instead of "24% (966MB/4GB)".
The balloon value is still tracked in memory.balloon for the frontend's
yellow balloon marker visualization, but no longer replaces the total.
Fixes#1070
Recovery notifications were bypassing the quiet hours check, causing
users to receive recovery alerts during their configured quiet hours
window even though the original "down" alerts were suppressed.
- Add ShouldSuppressResolvedNotification() to alert manager
- Check quiet hours before sending recovery notifications in monitor
- Recovery notifications now follow same suppression rules as alerts
This commit adds enterprise-grade reporting and audit capabilities:
Reporting:
- Refactored metrics store from internal/ to pkg/ for enterprise access
- Added pkg/reporting with shared interfaces for report generation
- Created API endpoint: GET /api/admin/reports/generate
- New ReportingPanel.tsx for PDF/CSV report configuration
Audit Webhooks:
- Extended pkg/audit with webhook URL management interface
- Added API endpoint: GET/POST /api/admin/webhooks/audit
- New AuditWebhookPanel.tsx for webhook configuration
- Updated Settings.tsx with Reporting and Webhooks tabs
Server Hardening:
- Enterprise hooks now execute outside mutex with panic recovery
- Removed dbPath from metrics Stats API to prevent path disclosure
- Added storage metrics persistence to polling loop
Documentation:
- Updated README.md feature table
- Updated docs/API.md with new endpoints
- Updated docs/PULSE_PRO.md with feature descriptions
- Updated docs/WEBHOOKS.md with audit webhooks section
- Replace barrel import in AuditLogPanel.tsx to fix ad-blocker crash
- Remove all Enterprise/Pro badges from nav and feature headers
- Simplify upgrade CTAs to clean 'Upgrade to Pro' links
- Update docs: PULSE_PRO.md, API.md, README.md, SECURITY.md
- Align terminology: single Pro tier, no separate Enterprise tier
Also includes prior refactoring:
- Move auth package to pkg/auth for enterprise reuse
- Export server functions for testability
- Stabilize CLI tests
1. Add IPOverride field to ClusterEndpoint struct
- Allows users to specify a custom IP that takes precedence over auto-discovered IPs
- Fixes#929 and #1066 where Pulse used internal cluster IPs instead of management IPs
- Added EffectiveIP() method to cleanly handle the override logic
2. Update connection code to use EffectiveIP()
- monitor.go: Use override when building endpoint URLs
- temperature_proxy.go: Use override for proxy connections
3. Add bare Windows EXE files to GitHub releases
- Fixes#1064 where LXC/barebone installs couldn't download Windows agents
- Modified build-release.sh to copy EXEs alongside ZIPs
- Added EXEs to checksum generation
1. Use correct mutex (diagMu) in cleanupDiagnosticSnapshots to prevent
"concurrent map iteration and map write" panics (Fixes#1063)
2. Use cluster name for storage instance comparison in UpdateStorageForInstance
to prevent storage duplication in clustered Proxmox setups (Fixes#1062)
3. Fix KUBECONFIG unbound variable error in install.sh by using ${KUBECONFIG:-}
default parameter expansion (Fixes#1065)
Shared storage was duplicating across polling cycles because the ID
included the node name of whichever node reported it first. When a
different node reported first on the next cycle, a new ID was created.
This fix updates the shared storage aggregation to use a consistent ID
format (instance-cluster-storageName) that doesn't include the node name.
Closes#1049. Thanks to @siccous for the report and initial investigation.
Allows administrators to create configuration profiles and assign them
to agents for centralized fleet management.
- Configuration profiles with customizable settings (Docker, K8s,
Proxmox monitoring, log level, reporting interval)
- Profile assignment to agents by ID
- Agent-side remote config client to fetch settings on startup
- Full CRUD API at /api/admin/profiles
- Settings UI panel in Settings → Agents → Agent Profiles
- Automatic cleanup of assignments when profiles are deleted
Implements server-side persistence for AI chat sessions, allowing users
to continue conversations across devices and browser sessions. Related
to #1059.
Backend:
- Add chat session CRUD API endpoints (GET/PUT/DELETE)
- Add persistence layer with per-user session storage
- Support session cleanup for old sessions (90 days)
- Multi-user support via auth context
Frontend:
- Rewrite aiChat store with server sync (debounced)
- Add session management UI (new conversation, switch, delete)
- Local storage as fallback/cache
- Initialize sync on app startup when AI is enabled
The storage deduplication logic only checked cluster config's Shared
flag, but this required the cluster config API call to succeed. When
the per-node storage API already returns shared=1 (as the user
verified), we should use that directly.
Now we check three sources for shared storage detection:
1. Per-node API shared flag (storage.Shared)
2. Cluster config shared flag (if available)
3. Storage type heuristics (NFS, RBD, PBS, etc.)
Related to #1049
Shared storages were appearing multiple times (once per node) because
the deduplication logic only checked the Proxmox `Shared` flag. Many
storage types are inherently cluster-wide but don't set this flag:
- RBD (Ceph block storage)
- CephFS
- PBS (Proxmox Backup Server)
- GlusterFS
- NFS
- CIFS/SMB
- iSCSI
Now we detect shared storage based on both the Shared flag AND the
storage type. Inherently shared storage types are deduplicated and
shown once with a "cluster" node designation.
Related to #1049
Changes:
1. Add MAX_POLL_TIMEOUT env var for large Proxmox clusters that need
more than 3 minutes for polling (default: 3m, minimum: 30s)
2. Handle external Ceph storage gracefully - don't mark nodes unhealthy
when Proxmox returns 'binary not installed' (e.g., for Ceph not
managed by Proxmox)
Related to #965
Previously, only PBS-type storages were queried when Active=0 because
querying inactive storage can return 500 errors. However, this caused
backups from datacenter backup tasks on shared storage (NFS, CIFS, etc.)
to not appear when the storage reported Active=0 on some nodes.
Now any storage with backup content is queried regardless of Active status.
If the storage is truly unavailable, GetStorageContent returns an error
which is already handled gracefully (logged and skipped).
Related to #1037
When a Proxmox cluster is discovered, Pulse now includes the user-provided
main host URL as a fallback endpoint. This handles scenarios where Proxmox
reports internal IPs that aren't reachable from Pulse's network (e.g.,
monitoring a remote cluster across different networks).
Previously, if all cluster endpoint IPs were unreachable, the connection
would fail with no fallback. Now the ClusterClient will fall back to the
main host URL, allowing Proxmox to route API calls internally.
Related to #1028
- Fix TestMonitor_PollGuestSnapshots_Coverage by correctly initializing State ID fields
- Improve PBS client to handle alternative datastore metric fields (total-space, etc.)
- Add comprehensive test coverage for PBS polling, auth failures, and datastore metrics
- Add various coverage tests for monitoring, alerts, and metadata handling
- Refactor Monitor to support better testing of client creation and auth handling
The mergeNVMeTempsIntoDisks and mergeHostAgentSMARTIntoDisks functions
require nodes to have LinkedHostAgentID populated to match disks with
host agent SMART data. However, the code was passing the local modelNodes
variable which doesn't have this field set - the linking happens inside
UpdateNodesForInstance which modifies the state's copy, not the local var.
Fixed by using currentState.Nodes (from GetSnapshot()) instead of
modelNodes/modelNodesCopy in both the skip-poll path and the background
goroutine. The state snapshot contains nodes with LinkedHostAgentID
already populated, allowing proper SMART data merging.
Related to #1014
When a container update command completed successfully, the server was
incorrectly returning shouldRemove=true, which caused the Docker host to
be removed and its API token revoked. This caused 401 Unauthorized errors
for subsequent agent reports.
The fix ensures shouldRemove is only true for "stop" commands, not for
"update_container" or "check_updates" commands.
Related to #1020
This resolves issues where snapshots/backups persist after deletion if the
Instance field didn't match the ID prefix (due to case changes, name changes, etc).
Now consistent with how VMs, Containers, Storage, etc. are filtered.
Also adds Instance field to BackupTask model for completeness.
Addresses #1009 (refs #991)
- Add persistent volume mounts for Go/npm caches (faster rebuilds)
- Add shell config with helpful aliases and custom prompt
- Add comprehensive devcontainer documentation
- Add pre-commit hooks for Go formatting and linting
- Use go-version-file in CI workflows instead of hardcoded versions
- Simplify docker compose commands with --wait flag
- Add gitignore entries for devcontainer auth files
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
pollGuestSnapshots was reading m.state.VMs and m.state.Containers while
only holding the Monitor's mutex (m.mu), not the State's internal mutex.
This caused a data race where VMs/containers could be modified by another
goroutine while being read, leading to stale or missing snapshot data.
Symptoms: Deleted snapshots persisting in UI, new snapshots not appearing,
only fixable by service restart.
Fix: Use GetSnapshot() which properly acquires State's mutex and returns
a consistent copy of the data.
Related to #991
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)
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.
Allows specifying which IP address the agent should report, useful for:
- Multi-homed systems with separate management networks
- Systems with private monitoring interfaces
- VPN/overlay network scenarios
Usage:
pulse-agent --report-ip 192.168.1.100
PULSE_REPORT_IP=192.168.1.100 pulse-agent
The GuestURL field was missing from NodeFrontend and its converter,
causing configured Guest URLs to be ignored when clicking on cluster
node names. The frontend would fall back to the auto-detected IP
instead of using the user-configured Guest URL.
Related to #940
- Add DELETE /api/agents/unregister endpoint for agent self-unregistration
- Agent now unregisters itself from Pulse server when uninstalled
- Add clarifying note in UnifiedAgents explaining linked agents behavior
- Linked agents are managed via their PVE node but this is now explained in UI
- Add LastSeen field to HostAgent model for better agent status tracking
When a host agent is deleted via the UI, the LinkedHostAgentID on any
PVE nodes that were linked to it was not being cleared. This caused
the "Agent" tag to persist in the UI after uninstalling the agent.
Related to #920
rcourtman