We've been talking about ToastStunt for several years now. It's a branch of Stunt, which itself is a rewrite/port of LambdaMOO to C++. It solves a bunch of problems for us, including the pending doom of running out of 32bit integers (still a ways off). It also has a bunch of improvements, additional builtin functions such as SQL integrations, https, curl, and maybe other things that make life better.
The caveat is that ToastStunt makes non reversible and non backwards compatible changes to our database. That means once we switch to it, we are pretty much on it forever. That's OK, because we've spent the last several years evaluating it. Our Dev Server (Boa) where we had rampage, has been running it the past year, proving that it is stable. We've worked out a number of kinks on Boa and during Rampage (thanks everyone for the help!).
Our current migration plan is still somewhat up in the air timeline wise but here are the general next steps:
1. hold another rampage
2. while doing the rampage, take the main MOO down temporarily and replace with toaststunt, so we can test our integrations (Slack, Jira, Grid, etc).
3. Bring main MOO back up on Production running LambdaMOO
4. post rampage, make a determination if we want to move forward
5. set a date (probably late December)
6. cut over production to ToastStunt
7. 2 week evaluation period -- rollback if we encounter big issues
The 2 week evaluation period will be where we really see if ToastStunt is ready for us long term. Rampage is great for proving the server can handle the load, but there are plenty of non-combat things that never happen during Rampage so we still expect to need to fix things during this period of time.
If we encounter big issues, we're going to turn off ToastStunt and go back to LambdaMOO. This means that if the 2 week evaluation period doesn't work out, we will be throwing away the RP that happened during that period of time, the same way we would if the server crashed. This probably sounds harsh and sucky and that's because it is, but because of the databases being incompatible, there isn't much we can do. We have evaluated as much as we can at this point (or we will have after the next Rampage). We're feeling pretty confident, but we also want to be clear about what our backup plan is.
If we rollback, none of the RP or accounts that were created on the MOO during that time will continue to exist. They'll have to be re-registered.
Once we have a date, we are asking that you plan around this 2 week period by not executing any major plots (perming someone, etc) that you aren't comfortable possibly being rolled back. That's right-- if someone is permed during this 2 weeks AND we rollback, they won't be permed anymore. It's happened before. If someone bombs a job interview and we rollback, they'll have a second chance. Please plan around this when we announce the date! Also, please don't plan to go crazy and do a bunch of unrealistic shit because we might roll back. Chances are good that we'll not be rolling back.
TLDR: During the 2 week period we are evaluating ToastStunt you should:
1. plan to avoid big plots coming to a head if you can help it
2. remember that this RP could need to be redone if we have to roll back
3. avoid perming people you aren't cool with coming back if we roll back
4. RP your character as you normally would -- this isn't a rampage.
I'm sure there will be a few questions. I'm happy to answer them. Let me preface it by saying the staff discussed a bunch of different ways we could approach this and we landed on this as the most clear, easiest to execute, and lowest impact on the coders, who will be focused on fixing any issues that come up as close to real time as possible.
That being said, if you have questions I'll be happy to answer them.