zhaohai666 opened a new pull request, #1038:
URL: https://github.com/apache/rocketmq-dashboard/pull/1038
# [RIP-1] Proxy config hot-reload, topology and health view (BASE-01 C)
## Background
The RocketMQ Studio `rocketmq-studio` branch exposes Proxy management only
through a thin
compatibility controller (`/api/proxy/*`) plus a `ProxyController` whose
single `POST /restart`
endpoint delegates to `ClusterService.restartProxy`, which throws `501 Not
implemented`. Proxy
nodes also never appear in cluster discovery because `RealClusterProvider`
always sets
`ClusterVO.proxies` to an empty list.
This change delivers the BASE-01 item C scope: concrete, unit-tested Proxy
management endpoints
(`list`, `config reload`, `topology`, `health`) and a dashboard view that
renders topology and
per-node health status with a working "Reload Config" action. The live Proxy
transport is modelled
as a **pluggable seam** — the default implementation is a no-op that returns
success so the command
path is fully exercisable end-to-end and in tests, while a real HTTP
transport can be enabled
without touching call sites.
**Base branch:** `upstream/rocketmq-studio` at `a8d263e`. The branch was
rebased onto the latest
`rocketmq-studio`; it has zero file overlap with the latest upstream commits
(including #1025), so
no merge conflict arises.
## Commits
```
3826050 feat: add Proxy config reload and topology endpoints (#RIP-1 T3)
6b7b058 feat: extend Proxy page with topology, health and config reload
(#RIP-1 T3)
305ff1b docs: add PR doc for Proxy config hot-reload, topology and health
view (#RIP-1 T3)
```
## Feature table
| Endpoint | Method | Purpose | Files | Tests |
| --- | --- | --- | --- | --- |
| `/api/proxies?clusterId=` | GET | List configured proxies (merged from
`ProxyAddressService`) | `ProxyController`, `ProxyService.listProxies` |
`ProxyControllerTest`, `ProxyServiceTest` |
| `/api/proxies/config/reload` | POST | Trigger proxy config hot-reload
(`{clusterId, addr}`) | `ProxyController`, `ProxyService.reloadProxyConfig`,
`ProxyConfigReloader` (+ `Noop`/`Http` impls) | `ProxyControllerTest`,
`ProxyServiceTest`, `ProxyConfigReloaderTest` |
| `/api/proxies/topology?clusterId=` | GET | Topology: proxies linked to
brokers / name servers | `ProxyController`, `ProxyService.getProxyTopology`,
`ProxyTopologyVO`/`ProxyTopologyNode` | `ProxyControllerTest`,
`ProxyServiceTest` |
| `/api/proxies/health?clusterId=&addr=` | GET | Health snapshot for one
proxy node | `ProxyController`, `ProxyService.getProxyHealth`, `ProxyHealthVO`
| `ProxyControllerTest`, `ProxyServiceTest` |
| `RealClusterProvider` discovery | n/a | Populate `ClusterVO.proxies` from
`ProxyAddressService` (read-only) | `RealClusterProvider` |
`RealClusterProviderTest` |
| Frontend: API + Proxy page | n/a | `listProxies`, `reloadProxyConfig`,
`getProxyTopology`, `getProxyHealth`; topology section, health tags, Reload
Config button | `web/src/api/proxy.ts`, `web/src/pages/studio/Proxy.tsx`,
`web/src/i18n/translations.ts` |
`web/src/pages/studio/__tests__/Proxy.test.tsx`, `web/src/api/proxy.test.ts` |
Input validation: `clusterId` is required for all four GET/POST management
endpoints (HTTP 400
otherwise); the reload body also requires `addr`.
## Testing
### Backend (offline, JDK 21)
```
cd server &&
JAVA_HOME=/Library/Java/JavaVirtualMachines/jdk-21.jdk/Contents/Home \
mvn -o test
-Dtest='ProxyControllerTest,ProxyServiceTest,ProxyConfigReloaderTest,RealClusterProviderTest,ProxyAddressServiceTest,ProxyCompatControllerTest'
```
Result: **32 tests, 32 passed** (ProxyControllerTest 11, ProxyServiceTest 4,
ProxyConfigReloaderTest 2, RealClusterProviderTest 3,
ProxyAddressServiceTest 9,
ProxyCompatControllerTest 3).
### Frontend
```
cd web && node_modules/.bin/tsc -b # exits 0
cd web && node_modules/.bin/vitest run
src/pages/studio/__tests__/Proxy.test.tsx src/api/proxy.test.ts
```
Result: **11 tests, 11 passed** (Proxy.test.tsx 6, proxy.test.ts 5); `tsc
-b` exits 0.
## Design notes
- **Pluggable reload seam.** `ProxyConfigReloader` is an interface.
`NoopProxyConfigReloader`
(`@Primary`) returns `success` without contacting a live proxy.
`HttpProxyConfigReloader` best-effort
`POST`s to `http://<addr>/admin/reloadConfig` and captures any failure in
the result VO (never
throws). Switching transports requires no change to `ProxyService` or the
controller.
- **Safe defaults.** Live metrics (`cpu`, `memory`, `tps`, `uptime`) and
broker/name-server links
are reported as `null` / empty when unavailable, so the UI renders
explicit "N/A" instead of
fabricated values. `RealClusterProvider` changes are read-only.
- **No unrelated refactors.** Only Proxy-related code and the Proxy
dashboard page were touched.
## Items not done (and why)
- **Live proxy transport wiring** is intentionally left as a seam (`Noop`
default). Pointing at a
real proxy admin endpoint requires deployment-specific addressing/config
that is out of scope for
this change; the `HttpProxyConfigReloader` implementation is ready and can
be enabled by qualifier.
- **`ProxyController` restart endpoint** remains a guarded seam delegating
to the existing
`ClusterService.restartProxy` (still `501`), per the instruction to keep
it but not implement live
restart here.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
