Self-host coroot
v1.29.2eBPF-powered APM and observability with automated insights
helm install -n coroot --create-namespace coroot-operator coroot/coroot-operator- Stars
- 8 006
- Ports
- 8080 (Coroot web UI and HTTP API listen address), 4317 (gRPC server for receiving OpenTelemetry traces and logs)
- Requires
- ClickHouse
ghcr.io/coroot/corootWhat is coroot?
Coroot gives SRE and DevOps teams a single platform covering the whole telemetry pipeline: collection, storage, visualization, alerting and diagnosis. A kernel-level eBPF agent captures signals automatically from every container on a node — no SDKs, restarts or manual setup — and turns them into a fully discovered service map, clustered log views, distributed traces and one-click CPU/memory profiles. The companion cluster agent watches databases found through the map (Postgres, MySQL, Redis, MongoDB and more), pulls app-level Go profiles, and discovers managed databases on AWS, GCP and Oracle Cloud; deployments and per-application cloud costs are tracked on their own, and reliability risks are surfaced proactively. An MCP endpoint backed by service-account API keys lets AI agents investigate incidents; the built-in inspections flag issues with no configuration, alongside SLO-based alerting to Slack, Teams, PagerDuty, Opsgenie or webhooks. Operators can deploy with a single Docker Compose command, or on Kubernetes via a Helm chart and an operator; Docker Swarm, ECS and plain Linux installs are also documented. Metrics live in Prometheus (or ClickHouse as a drop-in backend) while ClickHouse holds logs, traces and profiles; its own settings persist in SQLite, or PostgreSQL for HA setups. The Apache-2.0 Community Edition is free and unlimited on monitored infrastructure, while the paid Enterprise tier (from $1 per monitored CPU core per month) adds AI-driven root cause analysis, SSO and custom roles. It needs Linux kernel 4.16+, and Docker-in-Docker setups like Minikube and WSL1 are unsupported.
Should you self-host coroot?
coroot scores 69/100 on selfhost.directory (how scoring works) and ranks #4 of 9 in Application Performance Monitoring. It has 8 006 GitHub stars, +47 in the last 30 days.
It shipped 38 stable releases in the past year. The latest, v1.29.2, came out on 4 November 2026. New versions land roughly every 3 days.
Setup is moderate: expect some configuration beyond a single command. An official Docker image is available. It is Apache-2.0-licensed and written mostly in Go.
Its closest alternative on selfhost.directory is netdata (86/100, 81 345 stars). It scores higher overall, so compare the two before you pick. See coroot vs netdata side by side.
Features
- AI-powered root cause analysisEnterprise Edition feature (from $1 per monitored CPU core/month), not in the free Community Edition.
- Automated inspections
- Automatically discovered service map
- Cloud cost monitoring
- Continuous profiling
- Database Monitoring
- Deployment tracking
- Distributed Tracing
- Log monitoring with pattern clustering
- MCP server for AI agents
- Predefined dashboards
- Risk detection
- Role-based Access ControlCustom RBAC roles are an Enterprise Edition feature.
- SLO tracking and alerting
- Zero-instrumentation telemetry collection (eBPF)
coroot README
Collecting metrics, logs, and traces alone doesn't make your applications observable. Coroot turns that data into actionable insights for you!
More in Application Performance Monitoring
Durable ActorsSelf-hosted runtime for stateful serverless actors with durable persistence and serialized execution, replacing Cloudflare Durable Objects.
hypertraceSelf-hosted distributed tracing platform that ingests OpenTelemetry, Jaeger, and Zipkin data with service graphs, dashboards, and real-time metrics aggregation.
LinnixLinnix is a self-hosted eBPF agent that attributes stalls to specific processes and pods, finding noisy neighbors without SaaS.
LunalyticsA self-hosted, open-source monitor for sites, servers, and containers that handles multiple users, public status pages, and over a hundred alert channels.
MapleMaple is a self-hosted observability platform on OpenTelemetry and ClickHouse unifying traces, logs, metrics, session replay with an MCP server for AI agents.
measureAn open-source crash reporting and mobile performance monitoring platform with session replays, self-hosted via Docker Compose on a Linux VM.
n8nTraceObservability platform for n8n with execution analytics, Prometheus metrics, RBAC, and audit logging. Deployed as a hardened Docker container with PostgreSQL.
netdataNetdata is an open-source, self-hosted infrastructure monitoring platform with per-second metrics, automated dashboards, ML anomaly detection, and zero-config deployment.
netxmsSelf-hosted, open-source network monitoring with L2/L3 topology discovery, distributed agents, NXSL scripting, and no node limits on Linux.
NightOwl AgentOpen-source Laravel APM agent that captures Nightwatch telemetry into your own PostgreSQL database for full data ownership.
openobserveOpenObserve is a self-hosted, open-source observability platform unifying logs, metrics, traces, RUM, and AI observability with 140x lower storage costs.
pyroscopeOpen-source, self-hosted continuous profiling platform: ingest CPU/memory profiles via SDKs, Alloy or eBPF, store in object storage and query flame graphs in Grafana.




0 comments