Hi everyone,
I’ve just released nomad-state-metrics, a Prometheus exporter that generates metrics about the state of Nomad objects - jobs, allocations, nodes, deployments, and evaluations.
If you’re familiar with kube-state-metrics in the Kubernetes ecosystem, this fills the same gap for Nomad. Nomad’s built-in /v1/metrics endpoint covers agent-level telemetry, but there’s been no dedicated exporter for workload object state - things like job status, allocation health, restart counts, deployment progress, and blocked evaluations.
I built the original nomad-service-alerter (now archived) a few years ago (it was listed on the Nomad integrations page). nomad-state-metrics is a more complete approach - instead of baking in alerting logic, it emits standard Prometheus metrics so you can use your existing Prometheus + Alertmanager setup.
The repo includes example alerting rules and a Nomad job spec for deploying it on your cluster.
Would love feedback from anyone running Nomad in production.
Also, what metrics would be most useful for your setup?