The docker issue below starts on Jan 13, 2026.
It impact Symmetric Memory tests on H100 instances.
For example:
linux-jammy-cuda12.8-py3.10-gcc11-sm90-symm / test (h100-symm-mem, 1, 1, linux.aws.h100.4)
Login Succeeded
++ docker manifest inspect 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11-bfafaf0e518ed45e732eaef14740ce84d523e20a
++ jq '[.layers[].size, .config.size] | add / 1024 / 1024'
+ IMAGE_SIZE=15249.531907081604
+ echo 'Compressed size of image in MB: 15249.531907081604'
+ set -e
+ docker inspect --type=image 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11-bfafaf0e518ed45e732eaef14740ce84d523e20a
Compressed size of image in MB: 15249.531907081604
+ retry docker pull 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11-bfafaf0e518ed45e732eaef14740ce84d523e20a
+ docker pull 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11-bfafaf0e518ed45e732eaef14740ce84d523e20a
permission denied while trying to connect to the docker API at unix:///var/run/docker.sock
+ sleep 1
+ docker pull 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11-bfafaf0e518ed45e732eaef14740ce84d523e20a
permission denied while trying to connect to the docker API at unix:///var/run/docker.sock
+ sleep 2
+ docker pull 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11-bfafaf0e518ed45e732eaef14740ce84d523e20a
permission denied while trying to connect to the docker API at unix:///var/run/docker.sock
Error: Process completed with exit code 1.
cc @seemethere @malfet @pytorch/pytorch-dev-infra @atalman @fduwjj @dzmitry-huba @fegin
The docker issue below starts on Jan 13, 2026.
It impact Symmetric Memory tests on H100 instances.
For example:
linux-jammy-cuda12.8-py3.10-gcc11-sm90-symm / test (h100-symm-mem, 1, 1, linux.aws.h100.4)
cc @seemethere @malfet @pytorch/pytorch-dev-infra @atalman @fduwjj @dzmitry-huba @fegin