mengw15 opened a new issue, #6344:
URL: https://github.com/apache/texera/issues/6344

   ### Task Summary
   
   Add a boot smoke test for `ComputingUnitMaster` to the `amber-integration` 
job, mirroring the `texera-web` boot test added in #6319 (and the 
platform-service boot tests in #6274). Suggested by @Yicong-Huang in the #6319 
review.
   
   `ComputingUnitMaster` is the other server built from the amber dist 
(`bin/computing-unit-master`, port 8085); like `texera-web` it compiles and 
unit-tests clean but a runtime classpath/linkage regression (the #6204 class of 
bug) would only surface when its packaged launcher is actually started — which 
nothing in CI does today.
   
   ### Current state
   
   - #6319 boots `bin/texera-web-application` (port 8080) in 
`amber-integration` via `.github/scripts/smoke-boot.sh`, building the amber 
dist in-job (ubuntu-only, no artifact hand-off).
   - `bin/computing-unit-master` is produced by the same 
`WorkflowExecutionService/dist` but is not booted anywhere in CI.
   
   ### Suggested direction
   
   Add one more boot step in `amber-integration`, reusing the amber dist that 
#6319 already builds and unzips:
   
   ```yaml
   - name: Smoke-test computing-unit-master boots
     if: matrix.os == 'ubuntu-latest'
     run: .github/scripts/smoke-boot.sh 
"/tmp/dists/amber-*/bin/computing-unit-master" 8085
   ```
   
   It boots on **postgres only**, which `amber-integration` already provisions 
— verified against `ComputingUnitMaster.main`/`run`:
   
   - `clusterMode` defaults to `false` (no `--cluster` arg), so 
`AmberRuntime.startActorMaster(false)` binds a **local** Pekko actor system on 
`localhost:2552` and self-seeds a single-node cluster — no external infra, and 
the external `getNodeIpAddress` call (`checkip.amazonaws.com`) is only reached 
in cluster mode.
   - `ApplicationConfig.cleanupAllExecutionResults` defaults to `false` 
(asserted in `ApplicationConfigSpec`), so the boot-time result-storage cleanup 
(which would touch the iceberg/S3 result store) is skipped.
   - The recurring expired-result check is scheduled 2s post-boot but finds no 
executions in the fresh (empty) `texera_db`, so it does no result-storage work.
   - Config resolves the same way as `texera-web` — `Utils.amberHomePath` (a 
working-dir search for `amber`) → 
`amber/src/main/resources/computing-unit-master-config.yml`.
   
   ### Task Type
   
   - [ ] Refactor / Cleanup
   - [x] DevOps / Deployment / CI
   - [x] Testing / QA
   - [ ] Documentation
   - [ ] Performance
   - [ ] Other
   


-- 
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]

Reply via email to