I've just responded to this thread on our inability to delete virtual disks after a virtual machines were deleted. In our case, we started two VMs in an availability set in accordance with the SLA. (Actually 2 sets of servers in 2 availability sets and they're all broken now).
The VMs initially started okay, then appear to have restarted themselves and were stuck on a status of "Starting" for well over an hour. Apparently that is a known problem and the best solution is delete the VM and recreate it using the underlying disk/vhd.
That's precisely what we did. Now we can't recreate the VMs because the underlying disk is still attached to deleted VMs.
I've read various description regarding lease blobs not being released. Apparently others have experienced similar problems over the past year or two. They're all very nice solutions but the simple reality is THIS IS MICROSOFT FAULT and MICROSOFT SHOULD FIX IT because MICROSOFT IS BREACHING THEIR SLA.
Right now, our recommendation is to avoid Azure for production environments. It's proving to be too unreliable and it seems that their SLAs are nothing more than marketing hype.
I have attempted to contact support but apparently we need to purchase technical support for their faulty product. As I pointed out to their billing team, this is a BILLING ISSUE. If Microsoft is taking money from customers and not delivering the service they promised, it has become a billing issue.
If someone out there has has similar experiences and/or actually read the SLAs, I'd love your input. Rant over :)
Regards, Grant.