Blacklist test instead of stopping if bench fails.

[Menkib64]

There is fairly common case that a worker has GPU problems. GPU problems will cause a bench failure. Reporting of bench failure stops the test. This is different to how compile errors are handled using client blacklist. This change makes bench error handle the same as compile errors.

[Copilot]

Pull request overview

Aligns benchmark-failure handling with compile/build-failure handling by blacklisting the workload on the client rather than stopping the entire test on the server.

Changes:

• Server: client_bench_error no longer marks the test as finished; it only logs the bench error event.
• Client: on OpenBenchBadBenchException, the worker now appends the current test.id to config.blacklist before reporting the bench error.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
OpenBench/views.py	Removes server-side “stop test” behavior on bench error; keeps logging the event.
Client/worker.py	Adds the failing test id to the client blacklist on bench failure to avoid re-assignment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.