I know this is an old thread now, but rather than look at the technical limitations of the server/software, look at the lessons learned on how to handle a server crash in future since I doubt this one will be the last.
Instead of just having the server pop back on whenever it has finished rebooting (or whatever the term is), a GM could announce via forums, twitter, etc, that the server is ready and will be able to accept connections in 10 minutes time. This gives people time to call their friends, make announcements, and wake up from their desk. Once this 10 minutes is up, the GM hits the button and people start logging back in.
And then for the next 10 minutes, tidi should be at 0.01% to let the thousands of players to log back in. Once this 10 minutes is up, then the ticks start happening and the fight resumes as “normal” at 10% tidi or whatever.
It is a sucky idea but CCP does have to take responsibility for their server crashing.
It’s not an old thread, and it is still active. There were new posts just yesterday.
What you propose is fine I guess, except that you still have an unplayable game once more than X numbers of pilots are fighting in the same battle. I’d like to solve why more than X numbers of pilots can’t fight in the same battle, and why the server crashes happen in the first place.
The whole cluster crashes, or a node crash, make huge difference. When a node crashes only the grids hosted on this node have issues.
When the cluster crashes it refuses connections. When a node crashes, another node loads the system - and will crash too if same conditions are met. so while it is “loading up” the grids, you can’t be moved to the node /are stuck loging in. but people on other grids on the same node(ie if the node was hosting another system) should still be able to play.