[TRTLLM-10560][fix] Fix the time of pause() for overlap scheduler #10943
Conversation
Signed-off-by: Tailing Yuan <yuantailing@gmail.com>
|
/bot run |
📝 WalkthroughWalkthroughPyExecutor initialization parameters and control flow around request lifecycle management have been modified. The Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes 🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Fix all issues with AI agents
In `@tensorrt_llm/_torch/pyexecutor/py_executor.py`:
- Around line 1497-1498: The current pause flow calls _terminate_requests (which
uses _terminate_request) and thus removes result_wait_queues, breaking resumed
requests; add a new method (e.g., _release_resources_for_pause or
_pause_release_resources) that frees execution resources but does NOT delete
entries from result_wait_queues, update _pause_requests and all call sites
currently invoking _terminate_requests for pausing (including the calls near the
existing _pause_requests usage) to call this new release-for-pause path instead
of _terminate_requests, and ensure unit tests cover that paused requests retain
their result_wait_queue entries and can resume delivery.
🧹 Nitpick comments (1)
tensorrt_llm/_torch/pyexecutor/py_executor.py (1)
247-247: Name the new max_input_len sentinel for clarity.The hex literal is a magic value; consider promoting it to a constant with a short comment so intent is obvious to callers and future maintainers.
♻️ Suggested refactor
+MAX_INPUT_LEN_DEFAULT = 0x3FFFFFFF # effectively "no cap" for pause truncation ... - max_input_len: int = 0x3fffffff, + max_input_len: int = MAX_INPUT_LEN_DEFAULT,
|
PR_Github #33276 [ run ] triggered by Bot. Commit: |
|
PR_Github #33276 [ run ] completed with state |
|
/bot run |
|
PR_Github #33303 [ run ] triggered by Bot. Commit: |
|
PR_Github #33303 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #33330 [ run ] triggered by Bot. Commit: |
|
/bot run --disable-fail-fast |
|
PR_Github #33405 [ run ] triggered by Bot. Commit: |
|
PR_Github #33405 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #33432 [ run ] triggered by Bot. Commit: |
|
PR_Github #33432 [ run ] completed with state |
Signed-off-by: Tailing Yuan <yuantailing@gmail.com>
|
/bot run --disable-fail-fast |
|
PR_Github #33446 [ run ] triggered by Bot. Commit: |
|
PR_Github #33446 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #33462 [ run ] triggered by Bot. Commit: |
2 participants
Summary by CodeRabbit
Improvements
Refactor
✏️ Tip: You can customize this high-level summary in your review settings.
Description
Bug occurs if all the three conditions meet:
capacity_scheduler_policyisMAX_UTILIZATION, and pause happens;sampler_type: TRTLLMSampler.Whether pause happens depends on dataset, machine, and many other things. To make its reproduction easier, you can set a smaller
free_gpu_memory_fraction. Both mlperf dataset and random dataset can trigger the bug, if only the KV cache requirement is large enough to trigger pause.Reason:
pause(). Then, each paused request is regarded as new request whosemPromptLenincludes the original prompt and generated tokens.addNewToken(), which increases everyreqTokensby one. This includes the recent paused requests.reqTokens.size() == promptLencheck, wherereqTokenshas been increased by one. This causes assertion error.Compare to non-overlap scheduler:
By setting
disable_overlap_scheduler: true, I observed that pause happens andaddNewToken()is not called on paused requests, so there is no error.How to fix:
pause()to behindaddNewToken().reqTokensis truncated tomax_input_lenwhenpause()is called. If it's shorter than the length of the original request, some input is losing. So I change the default value ofmax_input_lento INT32 max to avoid truncation.Test Coverage
PR Checklist
Please review the following before submitting your PR:
PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.
GitHub Bot Help
/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...Provide a user friendly way for developers to interact with a Jenkins server.
Run
/bot [-h|--help]to print this help message.See details below for each supported subcommand.
Details
run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]Launch build/test pipelines. All previously running jobs will be killed.
--reuse-test (optional)pipeline-id(OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.--disable-reuse-test(OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.--disable-fail-fast(OPTIONAL) : Disable fail fast on build/tests/infra failures.--skip-test(OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.--stage-list "A10-PyTorch-1, xxx"(OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.--gpu-type "A30, H100_PCIe"(OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.--test-backend "pytorch, cpp"(OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.--only-multi-gpu-test(OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.--disable-multi-gpu-test(OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.--add-multi-gpu-test(OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.--post-merge(OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx"(OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".--detailed-log(OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.--debug(OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in thestage-listparameter to access the appropriate container environment. Note: Does NOT update GitHub check status.For guidance on mapping tests to stage names, see
docs/source/reference/ci-overview.mdand the
scripts/test_to_stage_mapping.pyhelper.kill
killKill all running builds associated with pull request.
skip
skip --comment COMMENTSkip testing for latest commit on pull request.
--comment "Reason for skipping build/test"is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.reuse-pipeline
reuse-pipelineReuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.