Well, I had the cluster running on the Linux server and was able to access it externally with DBeaver with the generated SSL certs. However, I started moving some things around and must have messed something up. I’ve tried removing all Cockroach data from my server to start from scratch, but no luck.
Whenever I try to reach my cluster externally from the IP and port 26257 it fails to connect. If I enter the cluster’s external IP and port in a browser (to see if the port is open), I get the following error:
<EXTERNAL IP> sent an invalid response. ERR_INVALID_HTTP_RESPONSE I can still reach the admin UI externally on port 8080 without any issues.
Now, I know it won’t show much when accessing the cluster address in a browser, but when it was working correctly it showed little boxes like this
 on the webpage and didn’t throw an error. It also shows these little boxes on the webpage when I run a browser window on the localhost.
The strange thing is I can reach the external cluster IP in the Microsoft Edge browser, and the
lsof command in Linux will show my IP connected to the cockroach db address/port. I tried flushing my local DNS records and reaching the cluster on other devices and IP addresses as well. I figured something strange may be happening on the Linux server, I made sure there was not additional instances running as well.
Attempt on new, second Linux server:
I then tried to start an additional, new Linux VM to narrow it down to some sort of configuration issue on the main server. I followed the same tutorial here, with the only change being adding
0.0.0.0 in the create-node certificate and start commands instead of
localhost. This is so that the cluster listens for external connections. But I can’t connect to this cluster node either, the connection attempt will timeout in DBeaver. The same
ERR_INVALID_HTTP_RESPONSE is shown on the webpage if I try to access the second, new cluster address in a browser window, but the Admin UI is reachable as usual.
The cluster is running, I’ve started it with the command:
cockroach start --certs-dir=certs --listen-addr=0.0.0.0
The console shows it’s running properly.
Any idea what could be causing these issues? If it was some sort of configuration issue on the main Linux server, the second test server should still work…