You may have noticed the messages that often appeared on the old WPMUDEV over the last three months. The messages anounced many failed attempts to upgrade this site and add new features. The reason that the prior attemps failed was that the original code that ran this website was developed rather quickly over two days, was not meant to be permanent, and was most *definetly* not scalable. Thus about 20 days ago it was decided that it was time to buckle down and rewrite WPMUDEV from the ground up. The new code took just under 100 hours to complete and was designed to be easily maintainable and upgradable. So, hopefully there will not be a need to perfom a complete rewrite in the near future. Although we are not quite finished with the API system, we went ahead and upgraded the live site to the new code for two reasons. The first being that the old and new db architectures were nowhere near compatible and we knew it would take several hours to transfer all the data. The second is that thursday night seems to be the period with the least amount of traffic according to google and waiting another week just wasn't feasible. Please feel free to ask any questions in the comments. Also, please note that invitations are no longer required to register for an account here at WPMUDEV.ORG!
Starting at around Midnight (in about 18 hours) we're going to be merging in about 35 bug fixes. Due to the large amount of code/files that will need to be updated, there is a possibility that this site might be down for a few hours.
A few bugs were fixed tonight and the frontpage received a bit of a face lift (a work in progress).
As many of you noticed, we had a bit of trouble keeping the site online last week. The reason for this was that a series of earthquakes damaged an underwater cable which affected all major upstream providers in Asia. At this time maintenance ships are still trying to repair the cable and there is no estimate as to when the repairs will be completed. Due to this news we will be shutting down all IDT servers in the Hong Kong data center which have been connected to the SAVVIS international ip network via a shared satellite connection since the incident. The reason for this is that we simply can't afford the $100 USD per day that the satellite connection costs. We have set up a temporary mirror in the SAVVIS data center near Boston, MA until repair work is finished and we can setup permanent mirrors. The downside to this move is that after reviewing the location of the underwater cable we realize that this will make the site unavailable to certain parts of Asia. For this we apologize but since 90% of the traffic for this site originates from North America and Europe it just doesn't make any sense economically to keep paying for the satellite connection.
All connectivity issues should now be resolved and the site has moved back to it's primary data center in Hong Kong. There are now two mirrors (Boston and Dallas, USA) that are synced every fifteen minutes.