Description
I've enabled a monitoring alert on non-staging droplets a while ago, for when the load average goes beyond 2x the CPU count (i.e. for droplets with 4 CPUs, when the load average goes exceeds 8.0, meaning more than ). I monitor the load average instead of the CPU usage as this more accurately reflects the backlog and load. High CPU usage isn't a problem per-se. See also https://en.wikipedia.org/wiki/Load_(computing).
Right now it only alerts to me directly, and unfortunately it seems DigitalOcean limits alert recipients to individually selected team members (no free-form input), and the team account, while having a registered an confirmed address, cannot be selected as receipient.
After a few months of little activity, it's been flapping recently due to some badly-behaving bots looking for nonsensical/non-existent URLs, thus missing the CDN cache and loading the droplet.
I'm planning to setup a rate limit in Cloudflare, that throttles how many (cache-miss) requests a given IP can make in a minute to the origin. I'm currently awaiting response on a request to enable the Rate Limit feature on our Cloudflare account.