as of 09/03/2026, I ran this code in AOU verily workbench, it failed with both 0.5.4/0.5.5.
%%bash --out test_ID
source ~/aou_dsub.bash #
aou_dsub
--name "${JOB_NAME}"
--logging "${WORKSPACE_BUCKET}/data/logging"
--image "${ARTIFACT_REGISTRY_DOCKER_REPO}/nvidia/cuda:12.9.2-cudnn-runtime-ubuntu24.04"
--boot-disk-size 80
--accelerator-type nvidia-tesla-t4
--accelerator-count 2
--output-recursive out_path="${WORKSPACE_BUCKET}/data/out_result/"
--command 'echo hello world > ${out_path}/test11.txt &&
nproc > ${out_path}/vm_cpu.txt &&
nvidia-smi > ${out_path}/gpu2.txt'
in contrast, 0.5.0 (after the patch of the google image issues) worked fine. this was found back one year ago and now with the 0.5.5, the issue is still there. can you please investigate and fix? Thanks.
as of 09/03/2026, I ran this code in AOU verily workbench, it failed with both 0.5.4/0.5.5.
%%bash --out test_ID
source ~/aou_dsub.bash #
aou_dsub
--name "${JOB_NAME}"
--logging "${WORKSPACE_BUCKET}/data/logging"
--image "${ARTIFACT_REGISTRY_DOCKER_REPO}/nvidia/cuda:12.9.2-cudnn-runtime-ubuntu24.04"
--boot-disk-size 80
--accelerator-type nvidia-tesla-t4
--accelerator-count 2
--output-recursive out_path="${WORKSPACE_BUCKET}/data/out_result/"
--command 'echo hello world > ${out_path}/test11.txt &&
nproc > ${out_path}/vm_cpu.txt &&
nvidia-smi > ${out_path}/gpu2.txt'
in contrast, 0.5.0 (after the patch of the google image issues) worked fine. this was found back one year ago and now with the 0.5.5, the issue is still there. can you please investigate and fix? Thanks.