Skip to content

test(dsv2lite): add retained HTTP soak gate#713

Open
CAICAIIs wants to merge 5 commits into
openinfer-project:mainfrom
CAICAIIs:feat/dsv2lite-http-soak-465
Open

test(dsv2lite): add retained HTTP soak gate#713
CAICAIIs wants to merge 5 commits into
openinfer-project:mainfrom
CAICAIIs:feat/dsv2lite-http-soak-465

Conversation

@CAICAIIs

Copy link
Copy Markdown
Collaborator

Summary

Adds a retained DeepSeek-V2-Lite HTTP soak runner for issue #465.

The new scripts/bench_dsv2lite_http_soak.py runner reuses the shared /v1/completions leaf benchmark, keeps per-bucket evidence, and writes backend-level plus combined host-staged/NCCL reports. The gate now fails on missing bucket coverage, failed leaf commands, request failures/timeouts, missing required trace coverage, failed clean follow-up, commit/provenance drift, or missing/generic backend runtime boundaries.

Closes #465.

Retained Evidence

Latest retained run:

  • commit: a5703d0424d917ce99b4bd8691b0b86eecde966f
  • model revision: 604d5664dddd88a0433dbae533b7fe9472482de0
  • hardware: 2x RTX 5090, CUDA 12.8
  • workload: prompt_words=64, max_tokens=64, greedy, ignore-EOS, concurrency 4,8, duration_s=120, bucket_s=60, num_requests=8
  • required trace coverage: 1.0

Results:

Backend Runtime boundary Completed Failed/timeouts Buckets / leafs Gate Artifact SHA-256
host-staged host-staged 112 0 / 0 4 / 14 passed 3fe4f163024602a51f10cac0c15cc24a5feffd92bbcb93091e175cd63e49bd33
NCCL NCCL 2.26.2, NCCL_IB_DISABLE=1, NCCL_P2P_DISABLE=1 128 0 / 0 4 / 16 passed 52c2a70d895fc0c080233961ea06824097fbf5b7241a0983e41c0ca6646e762f
combined host-staged + NCCL - - - passed 1c06e8825da70888277f1485f54d7f4fb9b2f61d617149d3ac7357cd5a03e7f1

The combined report passed child gates, commit consistency, model provenance consistency, and runtime-boundary retention.

Claim Boundary

This is short-shape sustained HTTP soak evidence for the named backend/runtime contracts. It does not claim production readiness, default-runtime NCCL coverage, long/mixed-prompt soak stability, multi-node recovery, vLLM parity, or a hard latency/memory drift budget.

Validation

python3 -m py_compile scripts/bench_http_common.py scripts/bench_http_serving.py scripts/bench_http_sweep.py scripts/bench_dsv2lite_http_slo.py scripts/bench_dsv2lite_http_soak.py
python3 -m unittest -v tests/test_bench_http_common.py tests/test_bench_http_serving.py tests/test_bench_http_sweep.py tests/test_bench_dsv2lite_http_slo.py tests/test_bench_dsv2lite_http_soak.py
cargo fmt --all --check
cargo metadata --locked --no-deps --format-version 1
git diff --check

python3 -m unittest passed 77 tests.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: eb4627294f

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

Comment thread scripts/bench_dsv2lite_http_soak.py
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

dsv2lite: add sustained HTTP soak benchmark for EP2 serving

1 participant