Worth verifying directly whether the issue is on Exoscale's end or something in the Cloudron configuration. The "code: undefined message: HTTP: undefined" error pattern specifically means the HTTP request to S3 is not completing at all - it's not getting a 4xx/5xx response from the server, it's getting no response or a connection-level failure.
A quick test from the Cloudron host itself would help clarify. Install s3cmd and test bucket listing directly:
s3cmd --host=sos-ch-gva-2.exo.io --host-bucket="%(bucket)s.sos-ch-gva-2.exo.io" ls s3://<your-bucket>/
If that also fails with a connection error, it confirms the issue is network-level between your Cloudron server and the Exoscale endpoint. Possible causes:
- DNS resolution failure for the S3 endpoint hostname
- A firewall change on your host or at Exoscale
- Their S3 service still not fully recovered in your region
Check if the hostname resolves: dig sos-ch-gva-2.exo.io
And test basic TCP connectivity: nc -zv sos-ch-gva-2.exo.io 443
If both resolve and connect fine, the problem is likely still at the API level on Exoscale's side. In that case, triggering a manual backup from the Cloudron dashboard and watching the logs in real-time will give you the full error chain: journalctl -fu box.service
If you need backups running today and Exoscale isn't cooperating, you can temporarily point the backup endpoint at a different S3-compatible service (Backblaze B2, Wasabi, etc.) to confirm your Cloudron setup itself is fine. That also rules out whether the issue is specifically with Exoscale or something in the Cloudron backup configuration.