We had an issue at 2pm today with our main deployment. All of our worker roles CPU's spiked up to 100% at the same time. Some of these worker roles are time based and run in the middle of the night, so we know it wasn't related to anything we were doing.
We also verifed in our logs we were not processing.
Specific times were 2:09 PM EST to 2:18 PM EST. At that time we scaled our web role to 3 instances. We then immediately saw the CPU drop on all the worker roles associated with that deployment as they restarted.
We wanted to let you know about it and find out what happened.