Grafana templates
Five drop-in Grafana dashboards keyed to z4j's metric names.
z4j ships five Grafana 10+ dashboards in deploy/grafana/: overview, tasks, agents/commands, notifications, and the scheduler companion. Each one is a standard import (Grafana, New, Import, pick datasource, save) or provisionable via a file-based dashboard provider for infra-as-code shops. Every brain dashboard exposes a $project template variable so multi-tenant deployments can scope panels to a single project without editing JSON. The split into five small dashboards over one mega-dashboard means each fits a single 1080p screen and matches the operator's natural drill path: overview -> per-area -> per-task / per-channel.
Ships with
- z4j-overview: agents online + RSS + DB pool + deadlocks + task throughput + queue depth
- z4j-tasks: failure rate, duration heatmap, top-10 failing / slow / retried task names
- z4j-agents: per-project agent + worker counts, command status mix, late-result counter, WS connections
- z4j-notifications: per-channel send rate, failure-rate table, cooldown skips, 24h channel mix donut
- z4j-scheduler: leader status, fire throughput, fire latency p50/p99 against the SLI budget
Highlights
- Schema-compatible with Grafana 10.4+; provisionable via file-based dashboards provider for GitOps
- Documented suggested alerts (zero agents, pool saturation, deadlocks, self-watch errors, channel failure rate, late command results)
- Every metric expression goes through `clamp_min` / `topk` so panels stay readable on cold deploys with sparse data