I’m using Proxmox Backup Server with a DreamHost S3 endpoint. Since upgrading to PBS 4 about a week ago, I’ve already hit TLS errors that stop backups. It happened within days.
When it fails, running openssl s_client against the endpoint shows a different SHA256 fingerprint than before. Updating the fingerprint in /etc/proxmox-backup/s3.cfg and reloading PBS fixes it — but only until the next change.
Example error:
POST /api2/json/access/ticket: 400 Bad Request: [client [::ffff:…]]
Plus TLS/HTTP errors in the S3 task logs.
What I’ve noticed:
The endpoint can resolve to multiple IPs, and some of them present different certs.
DreamHost appears to change certs or intermediates more often than expected (possibly load balancer/CDN behavior).
Possible solutions I’m considering:
Make PBS use the system CA store instead of a pinned cert.
If pinning is required, run a cron job to grab the current chain and refresh PBS trust automatically.
Force PBS to reload trust/chain data without manual restarts.
Has anyone else run into this with DreamHost S3? How did you make it stable?
When it fails, running openssl s_client against the endpoint shows a different SHA256 fingerprint than before. Updating the fingerprint in /etc/proxmox-backup/s3.cfg and reloading PBS fixes it — but only until the next change.
Example error:
POST /api2/json/access/ticket: 400 Bad Request: [client [::ffff:…]]
Plus TLS/HTTP errors in the S3 task logs.
What I’ve noticed:
The endpoint can resolve to multiple IPs, and some of them present different certs.
DreamHost appears to change certs or intermediates more often than expected (possibly load balancer/CDN behavior).
Possible solutions I’m considering:
Make PBS use the system CA store instead of a pinned cert.
If pinning is required, run a cron job to grab the current chain and refresh PBS trust automatically.
Force PBS to reload trust/chain data without manual restarts.
Has anyone else run into this with DreamHost S3? How did you make it stable?
Last edited: