Describe the bug
My issue stems from this commit which was included in the v2.19.0 release:
e23e44e
An issue pops up in our CI process due to the mounting of /opt/ml/metadata which is not a directory locally available in most cases and results in the following error:
ERROR: for tmpygfxtiwh_algo-1-amfg2_1 Cannot start service algo-1-amfg2: Mounts denied:
The path /opt/ml/metadata is not shared from the host and is not known to Docker.
Docker by default does not have permissions to mount to /opt/** on most local machines.
Ideally, this mount point would be configurable or mounted to the model_dir.
Expected behavior
Mount the metadata directory to the model_dir or allow support custom configuration.
System information
A description of your system. Please provide:
- SageMaker Python SDK version: 2.19.0
- Framework name (eg. PyTorch) or algorithm (eg. KMeans): Bring your own container
- Framework version: N/A
- Python version: 3.7.0
- CPU or GPU: CPU
- Custom Docker image (Y/N): Y
Additional context
Add any other context about the problem here.
Describe the bug
My issue stems from this commit which was included in the
v2.19.0release:e23e44e
An issue pops up in our CI process due to the mounting of
/opt/ml/metadatawhich is not a directory locally available in most cases and results in the following error:Docker by default does not have permissions to mount to
/opt/**on most local machines.Ideally, this mount point would be configurable or mounted to the
model_dir.Expected behavior
Mount the metadata directory to the
model_diror allow support custom configuration.System information
A description of your system. Please provide:
Additional context
Add any other context about the problem here.