Operations overview — Peren Docs

Check health before you change fleet state. Start with readiness and metrics, then use diagnostics, deployment health, and event streams.

Observe first

peren node health peren.toml
peren status peren.toml
peren diagnose peren.toml
peren deploy health peren.toml --service api
peren tail peren.toml --service api --level error

/healthz, /readyz, and /metrics are available on configured listeners. An operator may scrape /metrics with Prometheus. peren logs and peren tail require --service. See Observability.

Lifecycle commands that write registries

These commands update JSON under the data directory. They do not start a process and they do not move cells:

peren node drain — marks a node draining in the fleet registry
peren node remove — marks a node removed after a drain record exists, unless --force
peren deploy / peren rollback — update deployment generation records

Live refusal of new work on a running process is separate: POST /control/v1/node/drain on the peer listener. See Drain a node.

Backup and upgrade

peren backup peren.toml --output ./backup-data
peren upgrade check peren.toml
peren upgrade plan peren.toml

peren upgrade check tests the configured bucket. peren upgrade plan prints the steps and does not run them. See Plan upgrades.

When something is wrong

A ready listener can still have a failed provider or a drained cell. Start from the symptom, then check storage, peer readiness, and cell behavior. See Degraded operation.