[Etelis]

Related to #49238.

Bump the CUDA 12 NIXL pin to 1.5.0, which includes the upstream UCX teardown fix. #56907 already updated the standard requirements; CUDA 12 still uses 1.4.1. #59777 remains a draft for release-candidate validation; this uses the final release.

Tests

• CUDA 12 dependency resolution passed on x86_64 and aarch64.
• Pre-commit and git diff --check passed.
• GPU serving and prefill-restart passed

[claude]

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.

[robertgshaw2-redhat]

/ci run

[github-actions]

✅ Triggered Buildkite CI #93401 for commit a56fcd71c140.

[vllm-agent]

CI selector (shadow): 16 test steps (20 jobs) instead of 1 (1 jobs)

Shadow mode: this changes nothing about what CI runs. It shows what the evidence-based selector would pick for this PR, next to today's rules. How it works.

Feedback welcome: reply here if it would skip a step this change needs, or runs something unrelated.

steps (jobs)	Today's rules	Selector	Would skip	Would add
NVIDIA, CPU and others	1 (1)	16 (20)	1 (1)	16 (20)
AMD mirrors	1 (1)	2 (5)	1 (1)	2 (5)

Selector would run (16)

• crosslayer-kv-layout-distributed-nixlconnector-pd-accuracy-tests-4-gpus
• distributed-flashinfer-nixlconnector-pd-accuracy-4-gpus
• distributed-mooncakeconnector-pd-accuracy-4-gpus
• distributed-nixlconnector-pd-accuracy-4-gpus
• distributed-torchrun-examples-4-gpus
• dp-ep-distributed-nixlconnector-pd-accuracy-tests-4-gpus
• elastic-ep-scaling-test
• fault-tolerance-e2e-2xh100
• hybrid-ssm-nixlconnector-pd-accuracy-tests-4-gpus
• hybrid-ssm-nixlconnector-pd-prefix-cache-2-gpus
• kimi-k3-prefix-cache-4xb200 ×2
• multiconnector-nixl-offloading-pd-accuracy-2-gpus
• multiconnector-nixl-offloading-pd-edge-cases-2-gpus
• nixlconnector-pd-edge-cases-2-gpus
• push-nixlconnector-pp-prefill-pd-accuracy-4-gpus
• v1-kv-connectors ×4

Would skip (today's rules run them) (1)

• ray-dependency-compatibility-check

Would add (today's rules do not run them) (16)

• crosslayer-kv-layout-distributed-nixlconnector-pd-accuracy-tests-4-gpus (code map)
• distributed-flashinfer-nixlconnector-pd-accuracy-4-gpus (code map)
• distributed-mooncakeconnector-pd-accuracy-4-gpus (code map)
• distributed-nixlconnector-pd-accuracy-4-gpus (code map)
• distributed-torchrun-examples-4-gpus (code map)
• dp-ep-distributed-nixlconnector-pd-accuracy-tests-4-gpus (code map)
• elastic-ep-scaling-test (code map)
• fault-tolerance-e2e-2xh100 (code map)
• hybrid-ssm-nixlconnector-pd-accuracy-tests-4-gpus (code map)
• hybrid-ssm-nixlconnector-pd-prefix-cache-2-gpus (code map)
• kimi-k3-prefix-cache-4xb200 ×2 (code map)
• multiconnector-nixl-offloading-pd-accuracy-2-gpus (code map)
• multiconnector-nixl-offloading-pd-edge-cases-2-gpus (code map)
• nixlconnector-pd-edge-cases-2-gpus (code map)
• push-nixlconnector-pp-prefill-pd-accuracy-4-gpus (code map)
• v1-kv-connectors ×4 (code map)

AMD mirrors: would skip (1)

• ray-dependency-compatibility-check

AMD mirrors: would add (2)

• distributed-torchrun-examples-4-gpus (code map)
• v1-kv-connectors ×4 (code map)

_{1 changed files · base 74f4404f58 · head 3970cba2c6 · Python record: build 93039 at 1e5d0ea888 · kernel record: table f26d9f4 (build 94096), map f26d9f4 · not counted: 11 build steps, 5 A100 steps the generator no longer emits, 2 optional steps the selector would also run}

[Etelis]

/ci retry

[github-actions]

✅ Queued 3 failed job(s) for retry in Buildkite CI #93401.

[Etelis]

@Mergifyio refresh

[mergify]

refresh

✅ Pull request refreshed

[Etelis]

/ci run

[github-actions]

✅ Triggered Buildkite CI #94124 for commit 3970cba2c68e.