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 15, 07:15 UTC
[Build] Annotate wheel and container path for release workflow (#19162)
Signed-off-by: simon-mo <simon.mo@hey.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Simon Mo committed
Jun 4, 2025 at 23:24 UTC
da4038021480773e5a83b5f860681b49a7a0eafa
2 files changed
+49
-1
.buildkite/release-pipeline.yaml
+18
-1
@@ -1,5 +1,6 @@
1
steps:
2
- label: "Build wheel - CUDA 12.8"
3
+ id: build-wheel-cuda-12-8
4
agents:
5
queue: cpu_queue_postmerge
6
commands:
@@ -11,6 +12,7 @@ steps:
12
DOCKER_BUILDKIT: "1"
13
14
- label: "Build wheel - CUDA 12.6"
15
+ id: build-wheel-cuda-12-6
16
agents:
17
queue: cpu_queue_postmerge
18
commands:
@@ -28,6 +30,7 @@ steps:
30
31
- label: "Build wheel - CUDA 11.8"
32
# depends_on: block-build-cu118-wheel
33
+ id: build-wheel-cuda-11-8
34
agents:
35
queue: cpu_queue_postmerge
36
commands:
@@ -44,6 +47,7 @@ steps:
47
48
- label: "Build release image"
49
depends_on: block-release-image-build
50
+ id: build-release-image
51
agents:
52
queue: cpu_queue_postmerge
53
commands:
@@ -51,6 +55,18 @@ steps:
55
- "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.8.1 --tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT --target vllm-openai --progress plain -f docker/Dockerfile ."
56
- "docker push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT"
57
58
+ - label: "Annotate release workflow"
59
+ depends_on:
60
+ - build-release-image
61
+ - build-wheel-cuda-12-8
62
+ - build-wheel-cuda-12-6
63
+ - build-wheel-cuda-11-8
64
+ id: annotate-release-workflow
65
+ agents:
66
+ queue: cpu_queue_postmerge
67
+ commands:
68
+ - "bash .buildkite/scripts/annotate-release.sh"
69
+
70
- label: "Build and publish TPU release image"
71
depends_on: ~
72
if: build.env("NIGHTLY") == "1"
@@ -70,9 +86,10 @@ steps:
86
DOCKER_BUILDKIT: "1"
87
88
- input: "Provide Release version here"
89
+ id: input-release-version
90
fields:
91
- text: "What is the release version?"
75
- key: "release-version"
92
+ key: release-version
93
94
- block: "Build CPU release image"
95
key: block-cpu-release-image-build
.buildkite/scripts/annotate-release.sh
new
+31
@@ -0,0 +1,31 @@
1
+#!/bin/bash
2
+
3
+set -ex
4
+
5
+# Get release version and strip leading 'v' if present
6
+RELEASE_VERSION=$(buildkite-agent meta-data get release-version | sed 's/^v//')
7
+
8
+if [ -z "$RELEASE_VERSION" ]; then
9
+ echo "Error: RELEASE_VERSION is empty. 'release-version' metadata might not be set or is invalid."
10
+ exit 1
11
+fi
12
+
13
+buildkite-agent annotate --style 'info' --context 'release-workflow' << EOF
14
+To download the wheel:
15
+\`\`\`
16
+aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}/vllm-${RELEASE_VERSION}-cp38-abi3-manylinux1_x86_64.whl .
17
+aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu126/vllm-${RELEASE_VERSION}+cu126-cp38-abi3-manylinux1_x86_64.whl .
18
+aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu118/vllm-${RELEASE_VERSION}+cu118-cp38-abi3-manylinux1_x86_64.whl .
19
+\`\`\`
20
+
21
+To download and upload the image:
22
+
23
+\`\`\`
24
+docker pull public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}
25
+docker tag public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT} vllm/vllm-openai
26
+docker tag vllm/vllm-openai vllm/vllm-openai:latest
27
+docker tag vllm/vllm-openai vllm/vllm-openai:v${RELEASE_VERSION}
28
+docker push vllm/vllm-openai:latest
29
+docker push vllm/vllm-openai:v${RELEASE_VERSION}
30
+\`\`\`
31
+EOF
\ No newline at end of file