Many of us were affected by the Amazon EBS issues at the end of October 2012. If you had EC2 instances in us-east-1, you were likely affected by the issues. EBS volumes appeared “stuck,” snapshots would not complete, etc.
While the issues have been resolved (although we required some Amazon Support intervention for a few volumes), we have recently noticed what appear to be some vestigial issues related to the EBS outage.
The symptoms are, simply, that EC2 instances appear to be extremely slow. I/O is almost non-existent. Luckily, the fix is simple: perform a stop/start on the instance (not a restart). Your instance will be provisioned to new hardware, and you’ll have to ensure you account for a different IP address, but other than that, you’ll be back in business.
Of course- for next time- make sure that your instances are in multiple Availability Zones and Regions.
-Michael Klatsky, VP of Systems Administration