Basically, check the load average on the site whenever we receive a fetch request -- and reject if too high. This will only kick in when a remote site is fetching something from your site over ActivityPub. Everything else should still work normally...Basically, if we get a massive load spike, we're going to shut down access to a few of the most frequent remote-access GET endpoints on your site until it ends; but you should still be able to use the site (possibly degraded communications but more or less functional). That's the plan, and there's code to do it. It should just be a matter of finding the right value to cut it off at.
whether it's 20 or 30, nothing changes, the same old problem
uptime
This project focuses on stable and documented automated deployment for two of them - Hubzilla and Streams, including interoperability tests. This will support threadiverse standardization efforts, and help to bring features like group photoalbums and full channel portability between instances.
new instance, who dis?