The stacktrace has to do with Discourse not being able to talk to your SMTP server. Since this is done in a background job I don’t think this is what caused your downtime.
The monitoring agent not reporting for the first ~15 minutes of the downtime suggests that this might be a system issue rather than a Discourse issue.
Maybe @supermathie has ideas what to look next?