Step 3, webmail…
Ran through the setup steps, using both environment variables in the server and manual oauth directory registration through the console for webmail. Runs with node, gives the “is this your computer” page. Attempt to click sign in and it immediately bounces back with “Couldn’t reach the mail server. Try again in a moment.” Logging in systemd service shows “[ihasmail] could not start sign-in: fetch failed”.
Nothing in the nginx logs. This is behind two SSL terminations, instead of one like the console. (main SSL → proxies to secondary machine SSL → proxy to node)
Also, nothing that I can see in the inbuxa server logs.
… help?
Thanks for the detail. “fetch failed” means the webmail’s own Node process couldn’t connect to the mail server at all, which is why nginx and inbuxa logged nothing. On the webmail box, as the service user, could you run:
node -e 'fetch(process.argv[1]+"/.well-known/oauth-authorization-server").then(r=>console.log(r.status)).catch(e=>console.log(e.cause))' "$MAIL_SERVER_URL"
If it reports a certificate error, the second machine’s certificate isn’t trusted by Node. Add Environment=NODE_EXTRA_CA_CERTS=/path/to/your-ca.pem to the unit. If it’s ENOTFOUND or ECONNREFUSED, that URL doesn’t resolve or route from that host. Also check that the unit actually loads your env file.
We’ll make the log show the underlying reason too.
1 Like
Oooohhhh yeah that’ll do it. Got it!
Setting the cert did the job. … And now I have to go open a new support request for a woops.
all good, hope you notice, and I apologize it’s happening during your install, I am migrating off of MKdocs to my own CMS managed doc system. All built on HotDog-CMS https://hotdogcms.org/
1 Like
It looks really good! Took me about a second of “oh that’s different” before it became “okay cool” and I kept reading lol