Mail Delivery Fixed!
Apologies for the disruption
Some of you may have noticed the mailing lists on your projects were not delivering the mail sent to them over the past few weeks. We're really sorry about this! Users have been sending help emails about the problem, but because the problem was with mail delivery, the email addresses for the user group (email@example.com) and help (firstname.lastname@example.org) were also be affected, and the help requests weren't getting through. Once our volunteer sysadmin Ethan was informed about the problem, he immediately set out to get to the bottom of it. Hopefully, when you checked your email a couple of days ago, you would have noticed that all the undelivered mail sent over that time has now flooded your inbox. Eureka!
Ethan says the mail delivery problem was originally caused by "a disk filling up, which caused all kinds of lingering problems with both the mail sending and the mail delivering systems even after I cleared out the disk and rebooted crashed and hanging processes". He is now investigating moving the CoActivate mailserver to a separate server from the rest of the CoActivate system. He says this will "very likely improve deliverability / spam filtering errors and in fact probably improve delivery time and also result in minor speedups to the website too".
In the unlikely event of mail delivery problems with CoActivate in the future, there is a back-up help address on a separate service (email@example.com), which you can use to make sure the admins know about the problem. Yes, we know, the GoogleBeast! We will be looking into setting up a new backup address soon, using a service which respects its users' privacy and software freedom, such as Openmailbox.org or RiseUp.net.
Ethan has also been investigating why users have been having trouble using the 'Export' feature to take backups of their projects. "Back about a year and a half ago, I scaled up the hosting machine and set up four web-server processes behind a load balancer to get the most out of the server (that's why load times have been pretty fast since then, and - as far as i know at least - the proxy errors have more or less gone away for good). It's slightly unstable though, in that it's actually using a bit more hardware resources than are strictly available. So I also set up some scripts to reboot each of the four web-processes periodically throughout the day, one at a time, to free up memory and make sure nothing spins too far out of control. So the export will start, and then the server will happily kill it off before it finishes."
Doh! Rest assured that Ethan is working on a few possible ways of getting the 'Export' feature working again, and we'll let you know as soon as that happens. In the meantime, remember the old saying that the squeaky wheel gets oiled. If you notice anything on CoActivate not working as it should, or if you want to share your ideas about how it could be better, please email the user group or the help address, or send the admins a message using the contact form. If you use IRC, you can sometimes find Ethan idling in the #socialplanning channel, at irc.freenode.org.