Issue Station
Sign In
Tools hereToolbox

vllm - 💡(How to fix) Fix [CI Failure]: mi300_1: Acceptance Length Test (Large Models) [5 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
vllm-project/vllm#41321•Fetched 2026-05-01 05:34:15
View on GitHub
Comments
5
Participants
2
Timeline
17
Reactions
0
Author
AndreasKaratzasAndreasKaratzas
Participants
AndreasKaratzasAndreasKaratzasCaptainTimonCaptainTimon
Timeline (top)
commented ×5mentioned ×4subscribed ×4added_to_project_v2 ×2

Root Cause

  • Flaky test
  • Can reproduce locally
  • Caused by external libraries (e.g. bug in transformers)

Code Example

FAILED v1/spec_decode/test_acceptance_length.py::test_eagle3_acceptance_length[ROCM_ATTN-tp1-3-qwen3-30b-moe-vl-eagle3]
FAILED v1/spec_decode/test_acceptance_length.py::test_eagle3_acceptance_length[ROCM_AITER_UNIFIED_ATTN-tp1-3-qwen3-30b-moe-vl-eagle3]
FAILED v1/spec_decode/test_acceptance_length.py::test_eagle3_acceptance_length[TRITON_ATTN-tp1-3-qwen3-30b-moe-vl-eagle3]
RAW_BUFFERClick to expand / collapse

Name of failing test

(command rocm-smi || true) && export VLLM_TEST_GROUP_NAME=mi300_1-acceptance-length-test-large-models && export VLLM_ALLOW_DEPRECATED_BEAM_SEARCH=1 && cd /vllm-workspace/tests && export VLLM_ALLOW_INSECURE_SERIALIZATION=1 && pytest -v -s v1/spec_decode/test_acceptance_length.py -m slow_test

Basic information

  • Flaky test
  • Can reproduce locally
  • Caused by external libraries (e.g. bug in transformers)

🧪 Describe the failing test

FAILED v1/spec_decode/test_acceptance_length.py::test_eagle3_acceptance_length[ROCM_ATTN-tp1-3-qwen3-30b-moe-vl-eagle3]
FAILED v1/spec_decode/test_acceptance_length.py::test_eagle3_acceptance_length[ROCM_AITER_UNIFIED_ATTN-tp1-3-qwen3-30b-moe-vl-eagle3]
FAILED v1/spec_decode/test_acceptance_length.py::test_eagle3_acceptance_length[TRITON_ATTN-tp1-3-qwen3-30b-moe-vl-eagle3]

📝 History of failing test

  • Current streak start: 2026-04-25
  • First failure in 60d window: 2026-04-25
  • Last successful nightly: 2026-04-24
  • Break frequency (60d, pass↔fail flips): 1
  • Latest nightly date: 2026-04-29
  • Latest build(s): amd-ci #8058
  • Latest hardware status: mi300_1=fail

extent analysis

TL;DR

Investigate the recent changes in the test environment or codebase to identify the cause of the test failures.

Guidance

  • Review the test code in v1/spec_decode/test_acceptance_length.py to check for any recent changes that might be causing the failures.
  • Verify the hardware status of mi300_1 to ensure it is functioning correctly, as the latest hardware status indicates a failure.
  • Check the build logs of amd-ci #8058 for any errors or warnings that might be related to the test failures.
  • Run the test with a different hardware configuration or environment to isolate the issue.

Notes

The issue seems to be related to a specific test case and hardware configuration, so the solution might involve modifying the test code or adjusting the hardware settings.

Recommendation

Apply workaround: Investigate and address the hardware issue with mi300_1 as it is currently marked as failed, which might be causing the test failures.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#prompt template#agent execution#callback error#memory management#API rate limit

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

vllm - 💡(How to fix) Fix [CI Failure]: mi300_1: Acceptance Length Test (Large Models) [5 comments, 2 participants]