BaseAdminRoute

BaseAdminRoute gives trusted users a set of common administrative actions over the running service: clearing the cache, tailing logs live, reading release notes, and triggering a restart.

import { BaseAdminRoute, RouteDecorators } from "@rapidrest/service-core";
const { Route } = RouteDecorators;

@Route("/admin")
export class AdminRoute extends BaseAdminRoute {}

The CLI scaffold generates exactly this at src/routes/AdminRoute.ts, mounted at /admin.

Endpoints

Function	HTTP Method + Path	What it does
clearCache	GET /admin/clear-cache	Flushes the Redis-based second-level cache.
logs	UPGRADE /admin/logs	Opens a WebSocket that live-tails the service's log output.
getReleaseNotes	GET /admin/release-notes	Returns the service's release notes.
restart	GET /admin/restart	Signals the service to restart.

Every endpoint requires @Auth(["jwt"]), and clearCache, getReleaseNotes and restart additionally require the caller to hold one of the roles listed in trusted_roles. There is no way to reach them anonymously.

Configuration

conf.defaults({
  service_name: 'petstore',
  trusted_roles: ['admin'],
  datastores: {
    cache: {type: 'redis', host: 'localhost', port: 6379},
    logs: {type: 'redis', host: 'localhost', port: 6379},
  },
});

• clearCache only does anything useful if a cache datastore is configured. It scans and deletes keys matching db.cache.* on that connection.
• logs requires a logs datastore. On startup, BaseAdminRoute attaches a Winston transport (RedisTransport) that publishes every log line to a <service_name>-logs Redis channel; the WebSocket endpoint subscribes to that same channel and forwards messages to the connected client. If logs isn't configured, the route logs a warning at startup and the endpoint closes any connection attempt.
• restart publishes a RESTART message on a <service_name> (or service_admin, if service_name is unset) channel over the cache connection. BaseAdminRoute also subscribes to that same channel on startup, so any instance sharing that Redis connection (including the one that issued the request) will receive the signal and call process.kill(process.pid, "SIGINT") on itself. In a multi-instance deployment behind a shared Redis, a single /admin/restart call restarts every instance listening on that channel.

Locking it down further

Because AdminRoute is a normal class in your project, you can tighten access beyond trusted_roles. For example, mount it on a different base path that's only reachable from an internal network, or override trustedRoles to a narrower list:

@Route("/internal/admin")
export class AdminRoute extends BaseAdminRoute {
  protected trustedRoles: string[] = ['superadmin'];
}