We have a Windows Azure Web Role on two extra-small instances that has been running for weeks without problems. This morning, we unintentionally passed some spending limit, which resulted in Windows Azure shutting down our complete service, without any prior warning!
We removed the spending cap and began to re-deploy the Web Role, with the same codebase that has been running for weeks. To our astonishment, we got the deployment error
Validation Errors: Total requested resources are too large for the specified VM size.
We upgraded the deployment to two small instances instead of the extra-small instances, whereupon deployment was working again. Now, the web role is back in the web.
However, we still don’t understand why our deployment was suddenly too big for an extra-small instance. We didn’t change one bit since the last successful deployment to extra-small instances. We then tried to remove the deployment size by moving some files to Azure Storage, but even after reducing the package file by more than 1 MB, deployment still failed.
The cspkg file, the deployment package, is currently at 9'359 KB. If unzipped, the complete sitesroot folder’s size is 14 MB. Which is way below the 19'480 KB limit for the x-small instance.
Before we lose more time with trial-and-error, here’s my question: How exactly are those 19'480 KB calculated? Is it just the sitesroot folder, or the zipped package, or is it the sitesroot and approot folder together, or the whole unzipped package?
Thank you!
EDIT:
Could you verify if your local resources exceed 20 GB: