Actually, almost everything is moved over now except images, which are still doing their slow migration across the seas. All databases have now been baptized by the waters of the Atlantic and have made it over, and were restarted. You should be able to do anything that doesn't involve images on a much faster system today.
It wasn't a crash exactly, it's that mysql server on a central database would run but then refuse or drop connections to it. I think the call was made that instead of trying to debug or upgrade the server — which may well be a hardware failure in progress — it was better just to shut down the db servers and send the entire MariaDB directories. Considering that we didn't even bother to delete some binlogs, it went pretty fast. The more I think about the decision, the better it seems, since we could have had an equal amount of downtime trying to diagnose what went wrong, not moved any data in the process, and if it was a disk failure we could have lost data instead of uptime.
It wasn't a crash exactly, it's that mysql server on a central database would run but then refuse or drop connections to it. I think the call was made that instead of trying to debug or upgrade the server — which may well be a hardware failure in progress — it was better just to shut down the db servers and send the entire MariaDB directories. Considering that we didn't even bother to delete some binlogs, it went pretty fast. The more I think about the decision, the better it seems, since we could have had an equal amount of downtime trying to diagnose what went wrong, not moved any data in the process, and if it was a disk failure we could have lost data instead of uptime.
"Kitto daijoubu da yo." - Sakura Kinomoto