Persistent volume on S3 based storage

Currently trying to run a small cluster on scaleway instances. Got the cluster up and running but want to add some persistent storage. Scaleway offers object storage based on the S3 api. I tied to attach it as a storage class but cant figure out how to do this.

Can anyone help me to use this as a persistent volume within rancher?

Welcome @tjans,
am I right you want to use S3 as volume provisioner for a filesystem? Afaik this is not possible. S3 is an object storage. But you need a block storage for provide a filesystem. Maybe there are solutions out there that make this working, who knows?

Want kind of application do you want to deploy? Maybe community can help then.


@chris.ingenhaag thank you for your response.
Wanted to use the S3 to make persistend volumes to use on my pods. I want to deploy a small MariaDB/MySQL database to use for some personal projects so very little traffic and i’m looking for a cheap way to persist the data.

Ok, I think this is not possible afaik.

You may look if scaleway offers some block storage for this.


It is “possible” with things like but a bad idea in general… And a fantastically bad idea for a database in particular. The only way to update a file in S3 is to upload an entire new copy, and the whole database (or table with innodb_file_per_table). So you’d be rewriting the entire table/DB every time it changes.

As @chris.ingenhaag has said, you really need block storage for arbitrary volumes. Object storage is primarily for (relatively) high latency write-once/read-many high level API access to whole files. It’s not random-writable primary storage.

i knew that it would have bad performance but didn’t know that it had to rewrite the entire table/DB on every change. I’ll try to find something else for my persistent storage.