✅ Release Asset Validation (Post-Publish): PASSED

Assets were revalidated after publication due to a release edit.

Status: Live release assets re-validated ✅
Validated: 2025-12-23 23:23:53 UTC
Workflow: Validate Release Assets #68

Validation Summary

• All required assets present ✓
• Checksums verified ✓
• Version strings correct ✓
• Binary architectures validated ✓

Pulse v5.0.0

The biggest Pulse release yet. v5 brings AI-powered monitoring, native Kubernetes support, a unified agent architecture, and dozens of refinements based on community feedback.

Upgrade Guide: docs/UPGRADE_v5.md

Live Demo: demo.pulserelay.pro

What's New in v5

AI-Powered Monitoring

• AI Chat Assistant - Ask questions about your infrastructure in natural language. Bring your own API key (OpenAI, Anthropic, Google Gemini, Ollama, and more).
• AI Patrol (Pro) - Automated background health checks that run on a schedule and surface findings before they become outages.
• Real-time Anomaly Detection - Automatic baseline learning detects unusual CPU, memory, and disk patterns across your infrastructure.
• Alert Analysis - Optional AI analysis when alerts fire, providing context and suggested next steps.
• Cost Tracking - Track AI usage and costs per provider/model.

Kubernetes Support

• Native cluster monitoring via the unified agent with auto-detection.
• Resource visibility: Pods, Deployments, DaemonSets, StatefulSets, Services, Nodes, and Namespaces.
• Health status tracking for all workloads with init container awareness.
• Namespace filtering with wildcard support for include/exclude rules.
• Multi-cluster support - Monitor multiple Kubernetes clusters from one Pulse instance.

Unified Agent

• Single binary that auto-detects Docker, Kubernetes, and Proxmox environments.
• Simplified installation UI with step-by-step setup wizard.
• Connection URL override for complex network setups.
• Version mismatch detection with easy upgrade prompts.
• Linux agent archives included in releases for manual installs.

UI/UX Improvements

• Setup Wizard - Streamlined 3-step onboarding for new installs.
• Redesigned pages - Ceph, Replication, and Mail Gateway views completely rebuilt.
• PBS Datastores now included in the unified Storage view.
• Collapsible threshold sections for cleaner alert configuration.
• Guest tag filtering - Filter alerts by VM/container tags and name prefixes.
• Clickable Docker host names for easier navigation.
• Mobile improvements for better responsiveness on smaller screens.

Security & Operations

• OIDC refresh token support for long-lived sessions.
• Improved OIDC environment variable handling - Config now works even without oidc.enc file.
• Bootstrap token flow for secure first-time setup.
• Enhanced encryption safety preventing accidental key deletion.
• WebSocket connection stability improvements.

Docker & Container Improvements

• Proxmox 9.1+ OCI container support via the unified agent.
• Docker Swarm service monitoring improvements.
• Container health check failure detection in AI Patrol.
• Rate limiting fix for Docker overview page.

Performance

• Optimized metric sampling - Better balance between resolution and memory usage.
• Reduced log noise - Routine operations now use debug level logging.
• Faster initial patrols - Skip if one ran recently.
• RSS/Atom feed fallback for update checks when GitHub rate limits hit.

Pulse Pro

AI Patrol requires a Pulse Pro license. Free users get full access to the AI Chat assistant (BYOK) and all other monitoring features.

What Makes AI Patrol Different

Unlike chatting with a generic AI, Patrol runs automatically and sees your entire infrastructure at once - every node, VM, container, storage pool, backup job, and Kubernetes cluster. It's not just a static checklist; it's an LLM analyzing real-time data enriched with historical context.

Context Patrol sees (and general AI can't):

• Trend analysis - 24h and 7d patterns for CPU, memory, disk across all resources
• Learned baselines - Detects anomalies based on what's normal for your environment
• Capacity predictions - "Storage pool will be full in 12 days at current growth"
• Infrastructure changes - VM migrations, config changes, new deployments
• Resource correlations - Patterns across related resources (containers on same host)
• Your notes - User annotations explaining expected behavior ("runs hot for transcoding")
• Previous findings - Respects dismissed alerts, learns from past investigations

Examples of what Patrol catches:

• Storage pools approaching capacity (ZFS, LVM, Ceph)
• Backup jobs that silently failed or are overdue
• VMs that keep restart-looping without errors surfacing
• Clock drift across cluster nodes affecting Ceph/HA
• Container health check failures across Docker/Podman/Kubernetes
• Degraded Ceph OSDs or PGs
• Kubernetes pods stuck in Pending/CrashLoopBackOff
• PBS datastores filling up or jobs failing
• Resources with sustained high usage trending toward limits
• Any pattern the LLM identifies as unusual for your specific setup

Because it's an LLM with full context, Patrol catches issues that static threshold-based alerting misses - the subtle patterns and correlations that indicate trouble before metrics hit alert thresholds.

Try Pulse Pro → 14-day money-back guarantee.

Bug Fixes (Highlights)

• Fix zpool binary detection on TrueNAS SCALE (resolves ZFS storage over-reporting)
• Fix Guest URLs not being used for PVE/PBS/PMG navigation
• Fix deleted env-based API tokens reappearing after restart
• Fix thresholds set to 0 not properly disabling alerts
• Fix AI model selector dropdown overflow on small screens
• Fix backup matching to use instance+VMID (prevents cross-instance collisions)
• Fix host agent identity collisions with MAC fallback
• Fix WebSocket reconnection log flooding
• Fix patrol interval not obeying configured settings
• Fix race conditions in agent WebSocket message handling

Thank You

Special thanks to everyone who tested the release candidates and provided feedback in the v5.0 RC Discussion.

Full Changelog: v4.36.2...v5.0.0

Installation

Docker (recommended):

docker pull rcourtman/pulse:5.0.0

Docker Compose:
Update your docker-compose.yml to use rcourtman/pulse:5.0.0

See the Installation Guide for complete setup instructions.