Connect Selectel S3 to ClearML

In ClearML, you can connect Selectel S3 to store datasets, results, and experiment artifacts.

Open the ClearML configuration file clearml.conf. Update the api, aws and development: blocks:

api {
  ...
  files_server: s3://s3.<pool>.storage.selcloud.ru:443/<container_name>
  ...
}

...
sdk {
...
    aws {
      s3 {
        host: "s3.storage.selcloud.ru:443"
        region: "ru-1"
        key: "<access_key>"
        secret: "<secret_key>"
        use_credentials_chain: false
        credentials: [{
            bucket: "<container_name>"
            secure: true
        }]
      }
      boto3 {
        pool_connections: 512
        max_multipart_concurrency: 16
      }
    }
...
}
...

...
development {
  ...
  default_output_uri: "s3://s3.<pool>.storage.selcloud.ru:443/<container_name>/<path>"
  ...
}
...

Specify:

<container_name> — the name of the S3 bucket where datasets and artifacts will be stored. You can find the name in the control panel: in the top menu, click Products → S3 → Buckets;
<access_key> — the Access Key ID from an S3 key issued to the user;
<secret_key> — the Secret Access Key from an S3 key issued to the user.
<path> — the prefix in S3;
<pool> — the pool where S3 is located (for example, ru-1).

To upload datasets to ClearML Server, run a python script.

Sample script for loading a single dataset:

# Create dataset via Dataset class
from clearml import Dataset
dataset = Dataset.create(
    dataset_name="<dataset_name>",
    dataset_project="<project_name>",
    output_uri="s3://s3.storage.selcloud.ru:443/<container_name>/<path>",
)

# Add files to dataset
dataset.add_files(
    path="<local_path_to_dataset>",
)

# Upload dataset to ClearML Server
dataset.upload()

# Commit dataset changes
dataset.finalize()

Specify:

<dataset_name> — the name of the dataset, will be displayed in the WebApp;
<project_name> — the name of the project, will be displayed in the WebApp;
<container_name>/<path> — the S3 prefix from step 1;
<local_path_to_dataset> — the path to the dataset on the local machine</g.