[trtllm-agent]

Description

Temporarily waive 59 failed case(s) on DGX_H100. 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) reports unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_coexists_with_per_conversation_and_zero_interval broken on main for DGX_H100: failure signature tensorrt_llm.bindings.internal.batch_manager.kv_ca....hostoomerror: mmap failed: cannot allocate memory, onset 2026-11-05T08:25:22Z, latest failure 2026-11-05T19:23:27Z. Policy native_p3@1.0.0 selected it; https://nvbugs/7053309 tracks the fix. Scope DGX_H100 is the stage prefix that ran the failing test.

• Bug: https://nvbugs/7053309 (details, logs and 7-day test history are in the bug)
• Restore coverage when: Remove this line when https://nvbugs/7053309 is fixed or closed and the test passes on DGX_H100 post-merge; un-waive follows the bug, not the PR.
• Scope reviewed by: auto-waive at 2026-11-05T21:37:55.880696+00:00

Tests in this break event:

• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_coexists_with_per_conversation_and_zero_interval on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_entry_is_released_with_the_request on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_table_is_cleared_on_shutdown on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_hybrid_replay_buffers_size_by_tokens_per_gen_step[v2] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_replay_all_layer_commit_matches_contiguous_layout on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_replay_builds_affine_state_layout on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_ring_replay_allocates_rings_instead_of_double_buffer on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_allocates_mamba_state_and_dummy_indices on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_disagg_page_table_preserves_lifecycle_indices on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_disagg_page_table_uses_qwen3_next_conv_sections on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_dummy_indices_keep_cuda_buffer_address on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_disagg_page_table_covers_every_role on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_mla_reuses_prefix_after_two_chunk_prefill on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_mla_reuses_prefix_at_aligned_snapshot on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_free_resources_drops_stale_state_index_mapping on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fused_replay_promotion_matches_unfused_path on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_intermediate_states_size_by_tokens_per_gen_step on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_invalid_check_scans_distinct_attention_pools on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_mamba_state_views_use_logical_slots on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_metadata_supports_attention_only_pp_rank on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_nvfp4_page_table_omits_ssm_block_scales[auto] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_nvfp4_page_table_omits_ssm_block_scales[nvfp4] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_preserves_per_conversation_and_disables_periodic_snapshots on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_reserves_every_persistent_dummy_slot on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_shares_an_inflight_prefix_without_branch_snapshots[False] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_shares_an_inflight_prefix_without_branch_snapshots[True] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_static_dynamic_tree_capacity on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_supports_pure_mamba_pp_rank on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_swa_scratch_keeps_ssm_placeholder_rows on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_uses_upstream_min_snapshot_policy on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_waiter_points_after_history_advanced_commit on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_inforward_snapshot_commits_point_and_restores_prompt_end_state on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_inforward_snapshot_plan_drops_a_mark_in_a_mixed_batch on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_asymmetric_context_generation_transfer on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_count_in_the_per_slot_budget on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_replace_the_replay_caches[2] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_replace_the_replay_caches[None] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_acceptance_and_cuda_graph on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_allocates_logical_slot_caches on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_buffers_are_kvcm_state_buffers[Glm5NextCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_buffers_are_kvcm_state_buffers[KimiK3HybridCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[512-0] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[512-16777216] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[None-16777216] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_supports_attention_only_pp_rank on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_refuses_kv_pool_rebalance[no_replay] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[False-draft_records] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[False-replay_caches] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[True-draft_records] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[True-replay_caches] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[None-True-Glm5NextCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[None-True-KimiK3HybridCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-False-Glm5NextCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-False-KimiK3HybridCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-True-Glm5NextCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-True-KimiK3HybridCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[generation-True-Glm5NextCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[generation-True-KimiK3HybridCacheManager] on DGX_H100
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_state_index_setup_skips_unchanged_uploads on DGX_H100

PRs whose pre-merge runs hit this break (rebase or re-run after this merges):

• [TRTLLMINF-451][infra] Capture setup timeouts and defer unfinished cleanup until JUnit save #19333
• [https://nvbugs/6858496][fix] Fail disagg gen requests whose grammar matcher cannot be created #20903
• [TRTLLM-15944][feat] Enable SWA scratch reuse for Gemma3/4 with TRTLLM attention #20971
• [https://nvbugs/6891165][fix] Pass chat_template_kwargs in the request body of the JSON-schema serve example #21026
• [None][feat] Rubin merge back: MLA, DSpark, DWDP, Helix #21067
• [None][feat] Keep hybrid speculative replay state in private KV cache buffers #21324
• [None][feat] GEN slot preemption: pause decode requests in place to admit at-risk waiters #21334
• [TRTLLM-17272][feat] Add public LLM health queries and a private-access lint for the serving frontends #21393
• [None][fix] Fix draft KV over-allocation during chunked prefill #21546
• [TRTLLM-16624][fix] Fix Helix MLA decode kernel selection #21582
• [TRTLLM-16731][fix] Autotuner: skip redundant per-profile work for exclude_from_cache ops in tuning mode #21593
• [None][feat] Responses: name the engine request behind each reply #21602
• [None][perf] Build MSA fmha_sm100 JIT kernels for the visible GPUs' target only #21681

full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_coexists_with_per_conversation_and_zero_interval SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_entry_is_released_with_the_request SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_table_is_cleared_on_shutdown SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_hybrid_replay_buffers_size_by_tokens_per_gen_step[v2] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_replay_all_layer_commit_matches_contiguous_layout SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_replay_builds_affine_state_layout SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_ring_replay_allocates_rings_instead_of_double_buffer SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_allocates_mamba_state_and_dummy_indices SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_disagg_page_table_preserves_lifecycle_indices SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_disagg_page_table_uses_qwen3_next_conv_sections SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_dummy_indices_keep_cuda_buffer_address SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_disagg_page_table_covers_every_role SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_mla_reuses_prefix_after_two_chunk_prefill SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_mla_reuses_prefix_at_aligned_snapshot SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_free_resources_drops_stale_state_index_mapping SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fused_replay_promotion_matches_unfused_path SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_intermediate_states_size_by_tokens_per_gen_step SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_invalid_check_scans_distinct_attention_pools SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_mamba_state_views_use_logical_slots SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_metadata_supports_attention_only_pp_rank SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_nvfp4_page_table_omits_ssm_block_scales[auto] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_nvfp4_page_table_omits_ssm_block_scales[nvfp4] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_preserves_per_conversation_and_disables_periodic_snapshots SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_reserves_every_persistent_dummy_slot SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_shares_an_inflight_prefix_without_branch_snapshots[False] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_shares_an_inflight_prefix_without_branch_snapshots[True] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_static_dynamic_tree_capacity SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_supports_pure_mamba_pp_rank SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_swa_scratch_keeps_ssm_placeholder_rows SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_uses_upstream_min_snapshot_policy SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_waiter_points_after_history_advanced_commit SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_inforward_snapshot_commits_point_and_restores_prompt_end_state SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_inforward_snapshot_plan_drops_a_mark_in_a_mixed_batch SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_asymmetric_context_generation_transfer SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_count_in_the_per_slot_budget SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_replace_the_replay_caches[2] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_replace_the_replay_caches[None] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_acceptance_and_cuda_graph SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_allocates_logical_slot_caches SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_buffers_are_kvcm_state_buffers[Glm5NextCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_buffers_are_kvcm_state_buffers[KimiK3HybridCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[512-0] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[512-16777216] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[None-16777216] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_supports_attention_only_pp_rank SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_refuses_kv_pool_rebalance[no_replay] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[False-draft_records] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[False-replay_caches] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[True-draft_records] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[True-replay_caches] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[None-True-Glm5NextCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[None-True-KimiK3HybridCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-False-Glm5NextCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-False-KimiK3HybridCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-True-Glm5NextCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-True-KimiK3HybridCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[generation-True-Glm5NextCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[generation-True-KimiK3HybridCacheManager] SKIP (https://nvbugs/7053309)
full:DGX_H100/unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_state_index_setup_skips_unchanged_uploads SKIP (https://nvbugs/7053309)

What to check before merging:

• The bug's test-history links show the cases failing on main, not only on the trigger PR.
• The scope DGX_H100 matches the failing stage and no narrower waiver already exists.
• Nothing else is needed from you after merging: the bug owner removes these lines with the fix.

Provenance

• Detector native-p3/k3-w12h-fresh12h-nopass-hostclock/1.0.1, policy native_p3@1.0.0
• Run 91685f6e-5aff-47e8-b513-9a331b095319; snapshot SHA-256 6b9de299e0a60b16f56c9f5e3f10cc7bc240b9439bec64e644b42f4db79d9f9f; config SHA-256 622a43f13fb241c37a015e4ec2c812d7c8cecc1a04413b2618ca4ac119a77c81
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_coexists_with_per_conversation_and_zero_interval: incident 2266500f7a887c56ae08b6fdaff1c30b42f577b5db84f5a7aefe1b3b06f5104d, observation 197544f9fb8854112e720dee2b5846193e7932bfa3d59bd5b3212649a2b3f213
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_entry_is_released_with_the_request: incident 60d191ca6495b62c404024e200f6dddf627c6a7cf8a927842b6d0158a2753452, observation de5ff10e1c7a0c6049ccfc038cb87fdf3e8efa6f6c38f99ae93f61c7d9a907f5
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_branch_snapshot_table_is_cleared_on_shutdown: incident 71673bd8d5e6ee20f40407e27d4b476306a65dd91d797301d95c7577dad38850, observation 5e9afcaafffa6ec0b93a328ac0a9c0009b3a632935761958e74c4d13ac86613c
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_hybrid_replay_buffers_size_by_tokens_per_gen_step[v2]: incident a21c54abef2ddb696861d669ac71b1097b98258aeb6d55fdb4addf9b2e67ecee, observation 4b2e3cb60f521db27adbfab137c8ddcc3012723a76e9de655a75639eb1cfea73
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_replay_all_layer_commit_matches_contiguous_layout: incident febf22524b23069ef2d345d19479c4a835d62144525f16c8202c2eb82d70a919, observation 6ca1d516ab3979d15b18e0be400c62082e0a86c5b598c3687d744850318d8feb
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_replay_builds_affine_state_layout: incident a6ad94aab93ed1650b3b54dd8e105426161ff76d7294cf3f3c411f89c4ea5ff0, observation 451ecb359758a376e805d9395237bda2bbe959a7d2aa5b3466b8862cc6bc1cd9
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_gdn_ring_replay_allocates_rings_instead_of_double_buffer: incident 52127f5392e83db80a08f47ee4851f82768ad5166fccf032d3376d9d1c7965eb, observation b185c968204854107f324d3276f2481092d2a4fe8a5230c6e67e3e8e4838eaf0
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_allocates_mamba_state_and_dummy_indices: incident 1429ecde113e2dee470af7c22635103dcc66660fbb7b538082d8d52c0efc1d3a, observation 4e776718512c2c9fb49642493c04e5387d7ad64b5a823080c66e9725023a55f0
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_disagg_page_table_preserves_lifecycle_indices: incident 02c56bf596e048e7094d47408682a2b02343ce7d5a3f720dede63195550976c1, observation e38663c77094a1f88ecef5b51e8cb10ce6508132e4abd37238a0c01b82052d16
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_disagg_page_table_uses_qwen3_next_conv_sections: incident 5bdd44dbb1025074a7c696a02e2ec536a998731e40e1a09740d7890501cac610, observation 5df3ab7de077ed54c3731bd8e51a30f3615cc614326a195367c9ac47c75ec3b5
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_dummy_indices_keep_cuda_buffer_address: incident febf6c3c464091fbd56470ff79a6333788bc2331f5e0773a843236ebcb8d1eee, observation 4995ca5cf99f06441a66cb6a1cc52633714f6ecbd8b29c4945d2a240a3acc411
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_disagg_page_table_covers_every_role: incident 34b21eebbd9518fac963205874d722352d7a8fe90963760fa3bd502c782496a9, observation a362fbb284b8bcf0df816ec3fd4864c9181872319f34b0a1012271680fcdfb00
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_mla_reuses_prefix_after_two_chunk_prefill: incident 9f6ec232b613a2aeec0955328982085ebaea03c240d6404adf0cf228c81c05d1, observation a8038e72440da1254464bd247adea1741b38359130bae95c8b362ebaf6d9c765
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fp4_mla_reuses_prefix_at_aligned_snapshot: incident a64bac8cbada95322312f4be586768812a93d4462683431266719ba45a59e5aa, observation af5bd95ea1a3ccd0dfa07a47fb506690919b6c9b8f2086cf84d083a94c753301
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_free_resources_drops_stale_state_index_mapping: incident ba10c92e9b9d9a6a1d9b79564ad7855d71d313a1695b50145ac9742d70b8a9a0, observation b4291bfad96472e172c6903f38f5c6734a25cb1d215eeff6132c76f27b81ecd5
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_fused_replay_promotion_matches_unfused_path: incident 40df1aae746b19f6292715a1dfdbcbe104b3b3daffaca437d8233b29508528c1, observation c1ae4d1c9027aca3aec09cff9bbf1b8403951bce0219a9ec15b5189ddd6382c9
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_intermediate_states_size_by_tokens_per_gen_step: incident 09290da0457312b96408b01eb4d87bfa8946f2080387b24b14cfe8200b923b56, observation 76234fe2ca147a2a56836fa7a90a4f4d795d56f75d1c2e18d3e2d4c7a381c523
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_invalid_check_scans_distinct_attention_pools: incident c063b5833afb455c75b6140489bd43ffda4e4a665ac4e611a44e22ad99337ba2, observation b11c2dab2bbba6b4445aeb08e92e57074cfa64885022e1c02be8c21e633b7820
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_mamba_state_views_use_logical_slots: incident 599e9df57792a4c4c506d5f53267adaca958d948db2843523dc0319365d34335, observation 6a104caf9e9b3b948d879fc7c7470de5828be59ad490a85b224bf959783ccddb
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_metadata_supports_attention_only_pp_rank: incident 6a6fa078cb5d61cf083892c1cca78fd35f8ae9d5bf1df8b5052ddb697df80651, observation 5c36ae60dbe02227044604d9508c917deb8bd8327ad7a4f1ee4fb5319856934c
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_nvfp4_page_table_omits_ssm_block_scales[auto]: incident b97cf96668439453446a81deaff72cfa7956cb25ff047490d1ef97261256453a, observation 9457deb66ff684cd6c7a133fcf22836d89f180151cafc070fc3d48d6e9a3b80f
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_nvfp4_page_table_omits_ssm_block_scales[nvfp4]: incident 7f4dc0f9fed60d4f420952a8432da8c79dc4df31329532ce35396a47e77d9320, observation 591834c9ecde5e6235e37a3e6f8be9f2dbfac0acff46e758ee6c032e00e2faa5
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_preserves_per_conversation_and_disables_periodic_snapshots: incident 7667cba569d25fcb6c33c7e3ac2db16444367f3c378d0f40eb40567fc0dadd19, observation 26eb1ce1065c8c04e558e61efb9ea0460e813759461b262f488859209d7d601b
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_reserves_every_persistent_dummy_slot: incident 5e981de752a33f31a057b32b794ce4d17d7f0902c39d2e5bfcdc81f9d8333c03, observation 9c96182261c2e9dfa21fc9c09e72524dc69e5760016be87c3beb156e9c102c81
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_shares_an_inflight_prefix_without_branch_snapshots[False]: incident bea5b961d82f626e9600cd812f43412478b136441de2a214193391e10444cb97, observation 96876ca352074495776ed504ab1eb7d3848125df5d98067c6aef9a4c5042b5fa
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_shares_an_inflight_prefix_without_branch_snapshots[True]: incident 6c4da99cad8d7dd7d5b1d47a6214f5c3ca37869c53fa1d3a08bc0662d694bb45, observation 99e1f5bab99067e3dc0797145ca060a7765cfce2e45ceab35c9b867054688d47
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_static_dynamic_tree_capacity: incident 64ecd16d2258808fd2d3299d9ca375bf27a610a4495695dc228634d718325846, observation ab413ed033c8e049ef70ca2212f708d368ab742d044e7cd49dee3835a608f55e
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_supports_pure_mamba_pp_rank: incident f8a3f4292706e3183c6f2afb30e0514ad83bdd93a23ae45719cc2307e8221223, observation 13b3132ba6359f4442bc1349780d40c336db8c84dbced7e03f89fb3c7962a71f
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_swa_scratch_keeps_ssm_placeholder_rows: incident 6e3e18fd93cfe009c883c0e0a03f37e48e828347f421c0a87cdecdc92b3c57ad, observation 0b9501cda98928ff2988c2ec4a9fadfcabefd71b66ec73b3e40d46a3d37b33de
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_uses_upstream_min_snapshot_policy: incident 159b652fc3d16da50a43e6cd72eca6d7e2861064f720d4c0ab5d1e8f3808db19, observation 278a0f7b96ffe1f29a9e190aea841f071ae74c700e4e288d3229702081d86249
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_hybrid_waiter_points_after_history_advanced_commit: incident 4125461d902e6a777a9a014abb3976389e4a9a5f0a4054cddeea961713a658e2, observation 9487af52afb6500364e9fe4a9de1976f4c826e310e4768911d8a58aa62915007
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_inforward_snapshot_commits_point_and_restores_prompt_end_state: incident 9d9dac382a7978331914dd0dddd515be37c354535f522b6d9f58d3caa6738906, observation f186b8f5aee69e6663c4292178e564d8e9a65095b81371f5b08cc502fe2adeb1
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_inforward_snapshot_plan_drops_a_mark_in_a_mixed_batch: incident 6a2f70ceef515af377df6d7b6cf7265bf4cbac3d94d6eb8567e48abd34d02522, observation 1fd1ecf5b1a41fdad615af5048283af4e1b9c9dd2ec170bab709a75b63cf42c7
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_asymmetric_context_generation_transfer: incident deeb2eb3300b6a64a71d09e431a591e9da301bf619813ef5b288c679f6a91766, observation f7f68d0b03e578d6c745771d1479be56783881985e2c52ca04996a309f5268c9
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_count_in_the_per_slot_budget: incident 256556548b54197d35b2fd84556eb6102c31a54b9aaa64a11f2530b3ad9270f4, observation 9c097725669db79202e2dced5dc0a321e9cd52ff82bc87adcd16813d8f6ad517
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_replace_the_replay_caches[2]: incident 9ab3ea26ec8d4604f461c9cc33b4756a15c267f956736fbc225ed5ec4c633946, observation 465f16821695d09eddac1cf94a2bb87ca8380e66ff0c5ba430d0a7e8852cc001
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_draft_records_replace_the_replay_caches[None]: incident 0e911f6e7b996b5740557fa31a51d06ac98f3c7ad48cc22af8522564cca1661f, observation ee4d580fb8b2f48a67cfd5018a31a47e6a2a9e01f0d657b21cb160b3cfef97be
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_acceptance_and_cuda_graph: incident bd1f2d4d41e06fc1e3e54c7c775692d039e3e7fdc9d9264aed638a8e5986c4f9, observation 3adc81ae82af222a5cf11372516e51c0d33866eb9b34aeba38a1865293ed05c6
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_allocates_logical_slot_caches: incident b7a533aa173bd67b179cad1e71b9662c69397285db0137ef2ca887eead57213a, observation 58596566f678d00ca6881c313cab88a765c7a504861cb6f28143bcc86cebc612
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_buffers_are_kvcm_state_buffers[Glm5NextCacheManager]: incident f2faace1c7cb4652a362e166db00ed33cb46216c401b92ceee9b1fa4eb0d4f56, observation 3f5dd80e8db16127907b88e951faa39559b2b801d3dc7d5084d95945e417c720
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_buffers_are_kvcm_state_buffers[KimiK3HybridCacheManager]: incident b865584faa03e6471b320977b4cbe194572715bfa87ee55817f3feb608e5eef6, observation 48d452c84fab83c14f7e28148de7c07653d32ba8f7b275bab3dfe6e0fb0072ed
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[512-0]: incident 9fd81cb4ee8ae996adb26304f78b35bbbb71c6ac99ad94e162d5f5aa81e0a76b, observation f4e86de5e28c750480c7296b92de5221c187e5c9b37027ea9e0e8085d937a340
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[512-16777216]: incident 05c266c1ac6c36afb35a463a2f43237c4cfbfb039b7b2e5622514d572c734a9b, observation b4b940e0eb3086a46a42442831d9f7f779868238b546fa8f1ce960167a233c37
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_allocates_logical_slot_caches[None-16777216]: incident 109d37354f86009b7b1caef624a80a181deae6121123ffa883cc3285f5edb40b, observation 6b179e69a1367bd8ca1d8c993169b21b6c1d68565efd15abb7728fc2270fd2c4
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_policy_supports_attention_only_pp_rank: incident 174e6743d82e0b75236a31bfdef75082d6c3e4e5bbbc7d6b4682d0b71b32e8ae, observation b7224ad16ee1e82366eaf3ad89ede83f78e48697f0b9f973311450cb5d6823de
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_refuses_kv_pool_rebalance[no_replay]: incident 9c7cb6a227c3931c732b0764c4a226cd674a1bb373ae74d892605f1d4a280ee7, observation 935a8e01b4e8ccec2533d15b50dfb63dbcd5ff51363712266975087a420bb794
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[False-draft_records]: incident f0be2e0d834f6f4bd4e5fa5c7a492b9dd89f8da53aa29f9689bf4a7e3dacd39e, observation 10b3e847beef20f3802bbff6cbd3c4903889759754464dbc4899e13a19a06d84
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[False-replay_caches]: incident a53b451011dd54ead6541b74dacebf914b6bcd6c06860f302873368c90b10635, observation 6da5c7eaf2e1706a1d5d458640e95b1235c67460aa5dfbf17ed35a780702050c
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[True-draft_records]: incident ef1be4ffdebf3eb328220ab00902a71c3628ca4b2c76f5ece131c8c2458616b2, observation f8fabef7ce91ee1bca834a2ba5e34cb917abb9873fc76612a4de724298c822b0
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_replay_survives_cache_lifecycle[True-replay_caches]: incident 13653b325505cf31f5c6308a087a735e26df2c11d752c999cfd061d07aecef25, observation d9c31a3ca9e40884263d13afbb419fa8c6437f204ceb0c3b0dbfe62e4cee0656
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[None-True-Glm5NextCacheManager]: incident 0390e023e3bc3395dc2ff464e2e22b402379694d6af1257304fc8b04631f5d9e, observation 07ede59d7f1a1fb0919d167beb3cdfa5f09af99d1175c282ebb15ab319f87296
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[None-True-KimiK3HybridCacheManager]: incident 171f3020f23f3ab63b0a8ad6b2077ad1ca3ae0366150dced2b1728f966e33655, observation 3e8f96b86d36275f10ae2e7bb43ffc7be4c024e394a99d2ad36a559b2fcf7d6e
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-False-Glm5NextCacheManager]: incident 5d8fc5319e747bab938a973e51f788941a9af2d9c60dee1954251ffcd792b001, observation 2b5d206ace75f015e61b3bcde1fe3e2281b61ab370a1573a21285cf5a01a044b
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-False-KimiK3HybridCacheManager]: incident 8f6f1e1efdc8bb895ab8aec854c03dfe474cedce1bd05934d7036276f2cd4ee8, observation 1d7093109eaacc6a7dd3ca12f0a75bd987ee5d08a79ef36dfd1d94e8944d7b24
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-True-Glm5NextCacheManager]: incident cf68c1c947fd497ebb805c93f76f0d81addeeadfcf403e76fe7d2ef2aeabe88e, observation b44071e2cc4aa596c32182f4e2b4d508a8df12d4775ed514a3e679e79da05c40
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[context-True-KimiK3HybridCacheManager]: incident 2d3f0d85f3893df06f8291e1317d408729dea597b885d146fced450f850bec69, observation d093341e6d581ba54de8ed07570406ae67819613d5735d06ea90a48dbbfd72ee
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[generation-True-Glm5NextCacheManager]: incident d235813f021fdb3cd1d3384a93d4fe4f0523edb6cb16af92f66d0a3aad4fba29, observation a9f83af5e39e53516fc841cc2c0a60edcf3f0037398f9bbbee62697c5643da56
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_kda_worker_role_cache[generation-True-KimiK3HybridCacheManager]: incident 56ec47e76437b72494775641e8da12f9f56d320310c4a56ef5d1d8902fb4c1fa, observation 089b1825c1aaa440b5e1fb7f79c47d213d8637974b38785271e646ae4b4b5bac
• unittest/_torch/pyexecutor/kv_cache/test_mamba_cache_manager.py::test_v2_state_index_setup_skips_unchanged_uploads: incident e16d77a2f3f78ab3ecddbaddd053a06ec2e0d337121a129e794c06dd79e4fd50, observation 3c0f6057783fef081103d29ef219cb7d2fdb19e781d34c1a57e4794e37270429

Test 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

• Please check this after reviewing the above items as appropriate for this PR.

[karljang]

Closing: these failures were a DGX_H100 host-memory condition (host OOM / CUDA invalid device context across 16 nodes, 37 of 50 builds passing), not a code break, and the stage recovered on its own. No waiver needed. The linked NVBug is being closed as not a bug.