Repository navigation
[https://nvbugs/7113411][infra] Waive 1 failed case(s) on all platforms - #22539
trtllm-agent wants to merge 1 commit into
Conversation
Signed-off-by: trtllm-agent <296075020+trtllm-agent@users.noreply.github.com>
|
/bot skip --comment "auto-waive: waive-only change" |
|
No actionable comments were generated in the recent review. WalkthroughThe skip list now includes the SM107 Kimi K3 VL routing test, associated with NVBUG 7113411. ChangesRouting test skip
Estimated code review effort: 1 (Trivial) | ~3 minutes Change: Other Suggested reviewers: Merge Risk:
|
trtllm-agent-review
left a comment
There was a problem hiding this comment.
Reviewed the one-line waive. It takes effect and matches the failure: the test fails on main in every CPU-Generic run since #21996 with the _fp4_mla_generation_num_blocks_device ImportError, it is cpu_only and listed only in l0_cpu.yml, the line is sorted and passes the test-list validator, and the bug link and restore condition are present. Posted comments: 0 blockers, 0 possible blockers, 1 nice to have.
| unittest/_torch/attention/multi_gpu/test_mla_helix_context.py::test_k3_mla_helix_owned_prefill[tp1-cp4-gate-fp8-weights-bf16-kv] SKIP (https://nvbugs/7109810) | ||
| unittest/_torch/attention/sparse/csa2/test_module.py::test_fused_native_index_q_and_graph SKIP (https://nvbugs/7039506) | ||
| unittest/_torch/misc/test_guided_decoder_errors.py::test_executor_loop_fails_the_request_once_its_batch_is_synchronized SKIP (https://nvbugs/7068781) | ||
| unittest/_torch/modeling_v2/models/kimi_k3_vl/test_k3_sm107_routing.py::test_route_returns_the_target_class_only_on_sm107 SKIP (https://nvbugs/7113411) |
There was a problem hiding this comment.
[Nice to have] (R1: PR process) #22545 fixes this same failure. #22078 removed _fp4_mla_generation_num_blocks_device from fp4_mla/metadata.py, #21996's Kimi K3 code still imports it (_k3_fp4_mla_cache.py:58, k3r_fp4_mla/fp4_mla_kv_append.py:46), and #22545 moves the helper into _k3_fp4_mla_cache.py. Eight sibling auto-waive PRs (#22523, #22524, #22525, #22534, #22536, #22542, #22543, #22544) waive other tests that fail with the same ImportError, each under its own NVBug. None of these PRs or bugs is linked to the fix. If #22545 lands first and this merges afterwards, the line skips a passing test, which is the only CI check that K3 routing.route returns the SM107 target class. The restore condition keys only on nvbugs/7113411, so nothing prompts removing the line.
Suggest linking nvbugs/7113411 and the sibling bugs to #22545 and settling the merge order. If the fix lands first, close this PR. Otherwise, #22545 or the bug owner should delete this line together with the siblings' lines. Note that the PRs listed in the description need #22545 regardless, because the same ImportError aborts collection of three SM107 test directories, and no waive line can skip a collection error.
|
Automatically added "ci: full pre-merge approved" because this PR has satisfied the required GitHub review approvals. Unresolved review conversations and other required checks remain independent merge requirements. |
|
Closing: nvbugs/7113411 is closed, so this waive is no longer needed. If the failure comes back, the bot files a new bug and a new PR. |
Description
Temporarily waive 1 failed case(s) on
all platformswith unscoped line(s): the failure is a code-level error that fails on every platform, first seen on CPU. Opened automatically by auto-waive (draft; a human reviews and merges).The native main-break detector (
native-p3/k3-w12h-fresh12h-nopass-hostclock/1.0.1) reportsunittest/_torch/modeling_v2/models/kimi_k3_vl/test_k3_sm107_routing.py::test_route_returns_the_target_class_only_on_sm107broken on main for CPU: failure signatureimporterror: cannot import name _fp#_mla_generati...llm/_torch/attention/backends/fp#_mla/metadata.py), onset 2026-11-18T02:07:18Z, latest failure 2026-11-18T05:13:25Z. Policynative_p3@1.0.0selected it; https://nvbugs/7113411 tracks the fix. The line is unscoped: the failure is a code-level error that does not depend on the GPU (seen on CPU; stage prefixCPU), so every platform that runs the case fails it.Tests in this break event:
unittest/_torch/modeling_v2/models/kimi_k3_vl/test_k3_sm107_routing.py::test_route_returns_the_target_class_only_on_sm107on CPUPRs whose pre-merge runs hit this break (rebase or re-run after this merges):
What to check before merging:
all platformsmatches the failing stage and no narrower waiver already exists.Provenance
native-p3/k3-w12h-fresh12h-nopass-hostclock/1.0.1, policynative_p3@1.0.0b63a992e-b7c5-4815-92d3-fe7b0737fe16; snapshot SHA-2565fb2f3fdddfe9ce39251cec27d66dc9e9c21975354c430fd53248a3ea4b859d8; config SHA-256465a88e1b18cbe3245f999c7a82690481adb9fa9d03086d8e5e54ea20b3e7190unittest/_torch/modeling_v2/models/kimi_k3_vl/test_k3_sm107_routing.py::test_route_returns_the_target_class_only_on_sm107: incident9ac83197b44a5cd211800e64d1909e1669576b4df7fbf7345830d1c58524dfcc, observationfa25b86d133fa4ba146323147e318a33ccb1b89704debad7230af8d2dc99a8f0Test Coverage
Review the incident evidence and selector scope before merging. This proposal does not
establish that main is fixed. Confirm restoration separately after the corrective change.
PR Checklist
Dev Engineer Review
tests/integration/test_lists/waives.txt(+1/-0).SKIPfortest_route_returns_the_target_class_only_on_sm107. The failure is described as a code-level error. The skip hides that error until the bug is fixed.QA Engineer Review
tests/integration/test_lists/waives.txt.unittest/_torch/modeling_v2/models/kimi_k3_vl/test_k3_sm107_routing.py::test_route_returns_the_target_class_only_on_sm107 SKIP (https://nvbugs/7113411).needs follow-up. The waiver removes coverage. Follow-up should confirm the fix for NVBUG 7113411 and remove the entry once the test passes.Per-File QA Perspective
tests/integration/test_lists/waives.txt: Adds one skip waiver for the SM107 Kimi K3 VL routing test. QA should confirm the test passes on SM107 after the fix and then remove this line.