Show HN: Certificate Transparency Advisor – proactive alerting
github.com
github.com
Any feedback on the work thus far would be appreciated.
What kind of hardware/instance size do you actually need to run this and keep up?
However, I've found it surprisingly efficient for the type of service I'm running. It's currently running on a single Linode instance with 2G RAM. I've got load averages on the box of 0.05, so I think it's easily handling things. I have various parts of the service running in spawned threads, and I've tested that I can configure the timing to be more aggressive than it currently is it needed.
In terms of keeping up, I had the service shutdown for over 48 hours days whilst debugging a frustrating race condition in the PostgreSQL driver[1] and it managed to catch up in a few hours.
What has held up online registration is the place where load is in fact a significant issue - in testing, I set myself up monitoring *.cloudflaressl.com, which is known to rotate regularly. I nearly hit the Amazon SES free tier limit and gave myself an Office 365 inbox I can't open without crashing IE. Obviously I need to ensure users can't do the same.
[0] https://github.com/crtsh/libx509pq [1] https://github.com/epgsql/epgsql/issues/80