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 5, 18:30 UTC
[CI] Avoid concurrent docker pull in intel XPU CI runners to prevent rate limit issues (#38594)
Signed-off-by: wendyliu235 <wenjun.liu@intel.com>
wenjun liu committed
Mar 31, 2026 at 22:23 UTC
e8057c00bcaae00eb6790dba387d5d99bc0dc7f2
1 file changed
+21
-5
.buildkite/scripts/hardware_ci/run-intel-test.sh
+21
-5
@@ -239,13 +239,29 @@ fi
239
# --- Docker housekeeping ---
240
cleanup_docker
241
242
+aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin "$REGISTRY"
243
+aws ecr get-login-password --region us-east-1 | docker login --username AWS --password-stdin 936637512419.dkr.ecr.us-east-1.amazonaws.com
244
+
245
# --- Build or pull test image ---
243
-if [[ -n "${IMAGE_TAG_XPU:-}" ]]; then
244
- echo "Using prebuilt XPU image: ${IMAGE_TAG_XPU}"
245
- docker pull "${IMAGE_TAG_XPU}"
246
+IMAGE="${IMAGE_TAG_XPU:-${image_name}}"
247
+
248
+echo "Using image: ${IMAGE}"
249
+
250
+if docker image inspect "${IMAGE}" >/dev/null 2>&1; then
251
+ echo "Image already exists locally, skipping pull"
252
else
247
- echo "Using prebuilt XPU image: ${image_name}"
248
- docker pull "${image_name}"
253
+ echo "Image not found locally, waiting for lock..."
254
+
255
+ flock /tmp/docker-pull.lock bash -c "
256
+ if docker image inspect '${IMAGE}' >/dev/null 2>&1; then
257
+ echo 'Image already pulled by another runner'
258
+ else
259
+ echo 'Pulling image...'
260
+ timeout 900 docker pull '${IMAGE}'
261
+ fi
262
+ "
263
+
264
+ echo "Pull step completed"
265
fi
266
267
remove_docker_container() {