There are still a few lingering issues, which we've been working hard to fix and will be looking at again today. To those affected, I'm very sorry for the wait. Funds are safe, it's just a question of opening the pipes properly to let them flow normally.
You guys are still:
* Waiting until the last minute (after they've been waiting a week!) to tell people their accounts have been flagged as 'high risk'
* Not updating your blog with problems as they arise
* Being completely untransparent about the obviously-a-lie "24 hour rolling limit".
* Letting emails about people's already confirmed bitcoins being locked into your "wallet" go unanswered.
* Generally being completely opaque about both what is going on, and why people are getting flagged at the last minute.
My recommendation: charge more for your services and fucking hire some people to do some damage control--AND BE TRANSPARENT. It's great that you are in the reddit threads, and now on here.. but for the love of god, you are pissing EVERYONE off. There are people falling over each other to be your "customer support specialist" or whatever you were hiring for the other day; have them do some friggin' support.
Several pieces of the site scaled up about 10x in the past month or two, and this exposed some lingering performance problems. Somebody told us an analogy one time of rebuilding a car while trying to drive it down the freeway at 60MPH. That is a good approximation of what we're doing right now - in short, we are doing so carefully and with the help of great test coverage to make sure nothing breaks along the way.
Thank you your bearing with us. I wish I could say this is the last time we'll experience scaling problems in the next year, but it seems unlikely with the growth of bitcoin right now. If this seems like something interesting to work on, we are also hiring: https://coinbase.com/jobs
We'll post updates to Twitter as we come closer to a resolution. Thank you!
Which database systems are involved in the "database migration" described in the support center article? Please list at least the product names and versions.
What was the nature of the "refactor"? What programming languages and implementations were involved, for instance?
What exactly happened during the "database migration"? Was there any data loss?
What sort of backup infrastructure is in place?