Operate AISIX AI Gateway across its lifecycle — from pre-launch checks through day-to-day operation, incident recovery, and routing configuration.
Deployment work spans a lifecycle: decisions made before a gateway takes traffic, the operational habits that keep it healthy day to day, the response when something breaks, and the routing-layer configuration that shapes how requests reach it. The pages in this section are grouped by where they fall in that lifecycle, not alphabetically — start with the group that matches what you're doing.
The process-level settings a gateway needs before it can serve traffic — resource source, listeners, runtime dependencies, and the AISIX Enterprise connection.
Which network surfaces a gateway exposes — proxy, metrics/status, configuration store, control-plane connection — and the exposure policy each one needs.
The separate health and status endpoints for process liveness, traffic readiness, configuration state, and per-model health, and how to read them together.
Deploying AISIX behind a TLS-terminating egress device to govern traffic from IDEs and coding agents that must keep their official endpoints.
For fleet-wide monitoring — metrics, logs, and exporters across every gateway instance — see Observability. Troubleshooting is surfaced separately under Releases and support in the sidebar rather than duplicated here.
