Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
These are E2E tests that worked on a previous fork of mine, but with minor modification should work for any version.
Since these are copied from a previous PR, I will copy-paste that description now:
By default, these will check if the vLLM simulator is available in the local environment, if not - they are skipped and log a warning with the command required to build the vLLM simulator.
The vLLM simulator is built from source in the tests/e2e/vllm-sim.Dockerfile and extracted to bin/llm-d-inference-sim.
The tests use a class VllmSimServer that wraps the vLLM simulator with start and stop methods, then they call the guidellm command and assert the values generated in the output report.
The 2 test modules have varying run times:
test_successful_benchmark (2 tests) - 14.12s
test_max_error_benchmark (1 test) - 15.98s
The max error test takes longer since it allows guidellm to start sending requests before killing the vLLM sim, and then waits for the guidellm command to stop gracefully.
The successful tests share a single vLLM sim instance, hence are more efficient.
Details