StartTunnel 1.3.1 / StartOS 0.4.0.2 — clearnet URLs failing with ERR_SSL_PROTOCOL_ERROR

Hi,

I’ve just updated StartOS to 0.4.0.2 and StartTunnel to 1.3.1.

Since the update, services that were previously available via clearnet URLs are no longer accessible.

Before the update, my Start9 server appeared in the StartTunnel server list. After updating, it was no longer listed there and instead appeared as a client.

I can see that the published ports for the services appear to have been automatically re-added, with port 443 forwarding to port 443.

However, when I try to access any of the cleaner URLs, for example:

https://nextcloud.REDACTED.com

I get:

This site can’t provide a secure connection

nextcloud.REDACTED.com sent an invalid response.

ERR_SSL_PROTOCOL_ERROR

I’ve rebooted both the Start9 server and StartTunnel, but the issue remains.

This setup was working correctly before the StartOS / StartTunnel updates.

My domain DNS settings have not changed

Is there anything that changed in StartOS 0.4.0.2 or StartTunnel 1.3.1 that would require the tunnel or published ports to be reconfigured?

The Server→Client demotion is the odd piece here. It doesn’t match any know update issues. But it definitely needs to be resolved. Are you allowing the automatic port forwarding, or did you configure them manually.

Thanks for the reply.

I’ve tried both manually creating the port forwarding in starttunnel as well as letting the service interface set up automatically create the entries but neither work unfortunately.

I’ve managed to resolve the issue. It seems the problem may have been related to Let’s Encrypt pausing SSL certificate issuance for the services.

I had to manually unpause certificate issuance using the URL provided in the StartOS logs.

I also removed the server from Starttunnel and deleted the Starttunnel connection in Start9. I then created a new server in Starttunnel and added the new connection in Start9.

Everything appears to be working correctly now.

1 Like

Great! I’m glad you got it sorted. I’m not clear if the upgrade played any part in the issue, but we’ll have this as a data point if anyone else is seeing something similar.