Ok, I'm back from my academic sojourn and coordinating with the tech ops team to see what they've been up to wrt these speed/reconnect issues. I've been getting sporadic reports of some pretty intensive performance-tuning work being doing, but I don't know the details.
Today's developer scrum mostly involved synchronisation of the new German exitnode cluster - as I'm a bit out of the loop, I didn't pull the team aside to review these issues from this thread... but I'll get current on this in these early morning hours, so I can help coordinate the work of updating on current progress and so on.
I will say that a quick glimpse at the rough throughput graphs show major jumps in total network usage... like, major
jumps... since I last looked a week ago. No idea if this is enough to actually bump up against physical capacity constraints - I suppose it's possible, but seems unlikely.
Also, I heard one of the tech ops folks discussing some DHCP parameter tuning/bugs in the scrum last nite... that is likely related to the reconnect issue, and I'll dig into the logs of daily scrums to see how that all fits together.
Wow - good to be back!