Reputation: 81
Our azure worker roles are not autoscaling, it was only running 2 instances, even though cpu based autoscaling is configured for up to 6 instances and the dashboard cpu graph was around 90%. I then tried to turn off autoscaling and manually set the number of instances, but I got "Failed to change the scale setting for the production deployment of cloud service". When I manually reduced the number of instances to 1, it worked, but when I try to increase them again it failed. I then stopped the deployment, and started it again. I was then able to set the scale setting. It scaled up to 5 instances. The Current autoscale setting is minimum 8 instances, max 10, but it's stuck at 5 instances. Why is it failing to scale up? I have screenshots of everything.
Upvotes: 1
Views: 1869
Reputation: 81
I am pretty sure this was caused by running into the default Azure per subscription limit of 20 cores max. We have a few projects running, and the total cores of all deployments (including staging) in a subscription may not exceed 20. See limits here: http://azure.microsoft.com/en-us/documentation/articles/azure-subscription-service-limits/
To increase the limits, you have to contact Azure support. It's free: http://azure.microsoft.com/blog/2014/06/04/azure-limits-quotas-increase-requests/
Upvotes: 2