Operational pulse
Critical, warning, stale, and healthy computed from live metrics, plus counters for reporting, needs attention, stale, and network I/O.
Platform · Fleet Overview
Ten projects. Sixty machines. One page.
Per-project dashboards stop working the moment you have more than a few. The organization overview sits above all of them: every instance reporting in, every issue, every monitor, and every page that still needs a human.
Service facts
The yard, at a glance
How fresh the board is, and what counts as stale.
- 1 minrefresh cadence
- The instance table covers the last 30 minutes and the page refreshes itself every minute, with a manual refresh button.
- 3 minof silence = stale
- Stale outranks warning on purpose: a box that stopped talking is a bigger unknown than one running warm.
- 90days of incident history
- Ongoing incidents marked, and uptime excludes missed probes so downtime never reads as 100%.
One page, not ten
The fleet reads as one heartbeat.
Half-migrated fleets still read as one fleet: Kubernetes nodes group by cluster, plain hosts sit alongside under "Not in a Kubernetes cluster". Part of every Cloud plan, and MIT licensed with full parity when you self-host.
kubectl apply -f traceway-kubernetes.yamlA DaemonSet for nodes plus one small cluster Deployment; the token lives in a single Secret.
- Instances
- Discovered, not registeredAnything reporting server metrics under its own service.name; an instance exists exactly as long as it reports.
- Kubernetes
- DaemonSet + DeploymentNode CPU, memory, disk, and network, plus pod metrics, container stdout, cluster state, and events.
- Grouping
- k8s.cluster.nameFilter by cluster, namespace, node, or deployment; several clusters can share one project.
- Cross-project
- Monitors + issuesEvery monitor with 30-day uptime and average latency, and the last 24 hours of issues, each labelled with its project.
- Search
- Host to regionBy host, OS, architecture, cloud region, cluster, or node.
On this line
From fleet to one box.
Rows sort worst-first, and one click lands you in the dashboard for the machine that is actually on fire.
A whole cluster in one apply
Two OpenTelemetry Collector workloads cover node health, pod resource usage, container logs, cluster state, and events.
Triage without the project hop
Uptime, latency, and last incident across all projects, and recently active issues from every project in one feed.
URL-scoped drill-in
Selecting an instance opens its project's dashboards filtered to that machine over the last 30 minutes — shareable, and it survives every time-range change.
Access
No new powers.
The overview shows the organization's projects, which every member can already open, and respects per-project role overrides. Read-only members can view it; nothing is shared between separate organizations.
Next stop
Stop opening ten dashboards.
Fleet health, cross-project triage, Kubernetes, and on-call in the same tool that already holds your traces, logs, and errors.