Skip to main content
Version: Next


Loads from and saves to an Amazon S3 object.



s3 [--anonymous] <uri>


s3 [--anonymous] <uri>


The s3 loader connects to an S3 bucket to acquire raw bytes from an S3 object. The s3 saver writes bytes to an S3 object in an S3 bucket.

The connector tries to retrieve the appropriate credentials using AWS's default credentials provider chain.


Make sure to configure AWS credentials for the same user account that runs tenzir and tenzir-node. The AWS CLI creates configuration files for the current user under ~/.aws, which can only be read by the same user account.

The tenzir-node systemd unit by default creates a tenzir user and runs as that user, meaning that the AWS credentials must also be configured for that user. The directory ~/.aws must be readable for the tenzir user.

If a config file <prefix>/etc/tenzir/plugin/s3.yaml or ~/.config/tenzir/plugin/s3.yaml exists, it is always preferred over the default AWS credentials. The configuration file must have the following format:

access-key: your-access-key
secret-key: your-secret-key
session-token: your-session-token (optional)

<uri> (Loader, Saver)

The path to the S3 object.

The syntax is s3://[<access-key>:<secret-key>@]<bucket-name>/<full-path-to-object>(?<options>).

Options can be appended to the path as query parameters, as per Arrow:

For S3, the options that can be included in the URI as query parameters are region, scheme, endpoint_override, allow_bucket_creation, and allow_bucket_deletion.

--anonymous (Loader, Saver)

Ignore any predefined credentials and try to load/save with anonymous credentials.


Read CSV from an object obj.csv in the bucket examplebucket:

from s3 s3://examplebucket/obj.csv read csv

Read JSON from an object test.json in the bucket examplebucket, but using a different, S3-compatible endpoint:

from s3 s3://examplebucket/test.json?