Open Letter to Haxus from French Empire

[AnrDaemon]

I have much to say about the topic, but sine it's work time, I only highlight a few point now.

the first thing to do right now may be to create yourself a real testing environement

There IS a testing environment. The problem is that testing massively distributed system is not easy, and it may behave slightly different at a different scale. Ask CCP about it.

And please, please dammit, put HTTPS on your website... HTTP nowadays is lame

Please don't spread google-induced hysteria. No important information is transferred on the website.

We don't understand, at all, why you're not restarting servers when they're in a frozen state or if they crash. I mean, it's not like one of the first computer science thing you learn is "If it doesn't work, try rebooting".

Actually, you'd be surprised how young this motto it. "Simply rebooting" only helps when you KNOW the program should work fine, but somehow it does not. And you reboot to free the environment from external factors. I.e. isolate the testing case.
In cluster, like SoH, servers are not isolated, and the game itself IS an external factor. Simply rebooting the crashed server would either cause it to crash again, or propagate that same crash to other servers. At the least. At worst, it would cause corruption in the environment (means, destroying the game itself).

State of the art software developement nowadays uses this everywhere.

No? Can you name a few?

I don't know for games for sure because I don't work on that domain, but I don't see at all why it couldn't be applied there too.

See above.

When a freeze or a crash happens, you save the logs and you restart, that's all.

Restart what? It's not just a single governing server, that does everything. It's not Lineage2, where you have 3 server. World, NPC AI and logging. It's a part of whole system. A very SMALL part.

It can be automated.

Yes. It can be.

Or you could move stuck avatars into a working server

Into a random system, where they have no environment to replicate the crash state, so it does not happen again.

a sort of a rollback

What rollback?

I guess it should be obvious.

No, it's not.

Gameplay and implementing functionalities

Here I would agree on one thing: The game now is both more complex to play and less challenging as a game.

About that, it seems your decay rules are getting harder and harder with each year passing.

I'd say, the problem is not decay rules per se, but a management overhead to maintain the colonies.