Uploading data with AWS tools

The profile from Reading data with AWS tools also writes:

# one file
aws s3 cp data.parquet s3://my-org/my-data/data.parquet --profile source-coop

# a folder, copying only new and changed files
aws s3 sync ./output s3://my-org/my-data/output/ --profile source-coop

# remove an object
aws s3 rm s3://my-org/my-data/old.parquet --profile source-coop

The data proxy accepts a write when all of these are true. Otherwise it answers AccessDenied.

• You may write to the product: you are a member of it, or of its organization, as write_data, maintainers or owners (see Members).
• Its data connection accepts writes. A read-only connection refuses them, whoever writes.
• Your role writes. The default role does. ReadOnly does not.

Name the object, or end the path with /. aws s3 cp data.parquet s3://my-org/my-data has no object key, and the proxy refuses it with Missing object key.

To upload from a GitHub Actions workflow or a script, sign in as a service account with write access to the product (see Access to products). service-account usage prints a workflow that signs in (see Use a service account). Replace its last step, aws s3 ls, with your aws s3 sync.