Day-2: what to watch, how to upgrade, what a backup must contain. Each page
here owns the install-method-agnostic contract; the exact commands for your
install shape live in the installation guides,
and the compose-shaped runbook an agent can execute is
agent/operate.md.
| I want to… | Page |
|---|---|
| Check it's healthy, watch logs, see what agents did | Monitoring & logs |
| Move to a new release | Upgrades |
| Survive a disk failure or a bad migration | Backup & restore |
| Fix something that's broken | Troubleshooting — the common symptoms, each linking to its fix |
One recurring human task lives in the UI rather than these pages: agents ask
for access by filing a request, and an operator approves or fulfils it (and
enters any credential secret) in the console at /app/access-requests —
find the request by the id the agent reports. The flow end to end:
first brokered call.
Quick health check:
jenticctl status # install, server, identity health
curl -fsS http://127.0.0.1:8000/health # app answers (liveness only)The 127.0.0.1:8000/8100 ports fit local Docker-shaped installs; Helm and
remote installs need a port-forward or their own host. What the health
endpoints do and don't tell you (they stay green with the database down),
plus the broker's /ready probe: Monitoring & logs.