A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 2, 11:00 UTC
[CI] Separate non-root smoke tests from image build step (#43712)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Kevin H. Luu committed
May 28, 2026 at 23:34 UTC
648c3ebee673cf0ece078a94d07a24ef0c7e63ca
1 file changed
+14
-2
.buildkite/image_build/image_build.yaml
+14
-2
@@ -6,14 +6,26 @@ steps:
6
timeout_in_minutes: 600
7
commands:
8
- if [[ "$BUILDKITE_BRANCH" == "main" ]]; then .buildkite/image_build/image_build.sh $REGISTRY $REPO $BUILDKITE_COMMIT $BRANCH $IMAGE_TAG $IMAGE_TAG_LATEST; else .buildkite/image_build/image_build.sh $REGISTRY $REPO $BUILDKITE_COMMIT $BRANCH $IMAGE_TAG; fi
9
- # Non-root smoke 1: the default (root) image must still be importable
9
+ retry:
10
+ automatic:
11
+ - exit_status: -1 # Agent was lost
12
+ limit: 2
13
+ - exit_status: -10 # Agent was lost
14
+ limit: 2
15
+
16
+ - label: ":docker: :smoking: Non-root smoke tests"
17
+ key: image-smoke-test
18
+ depends_on:
19
+ - image-build
20
+ commands:
21
+ # Smoke 1: the default (root) image must still be importable
22
# under a non-root UID via `--user 2000:0`. Validates the `vllm` passwd
23
# entry + group-0-writable /home/vllm + uv path cleanup from #31959.
24
# Uses `import vllm` rather than `vllm serve --help` because the latter
25
# instantiates `VllmConfig` which requires a GPU attached to the
26
# container.
27
- docker run --rm --user 2000:0 --entrypoint python3 "$IMAGE_TAG" -c "import vllm; print(vllm.__version__)"
16
- # Non-root smoke 2: assert the non-root enabling invariants are baked
28
+ # Smoke 2: assert the non-root enabling invariants are baked
29
# into the image. Runs as UID 2000:0 via a shell so we can verify
30
# filesystem perms + passwd/group file state + wrapper presence without
31
# triggering vLLM's GPU-requiring config-init path. The opt-in