Provide an hours notice for any Underlying Host reboot that isn't Automatic/service healing
For any underlying host reboots that happens on Azure there are always reboots that Azure can give prior notice about, it could be a hotfix you are applying to your host or you found an issue with the host and you need to reboot. If you gave us an hours notice instead of just rebooting hosts and in-turn VMs as and when you want, then we could have ramped up to 3 or more VMs to ensure we retained the high availability. The small increase in cost for this we are prepared for. We are not prepared for random (to us) reboots of components of our stack and especially not during peak hours of business.
That's will be nice.
Mark mauthe commented
sounds like a good idea
Scott St. Laurent commented
We just had a customer suffer random reboots on two different VMs on four different occasions over two weeks. We were notified after the fact that the host took a "self-healing" action on those occasions. While it's almost impossible to send notifications out when a hardware failure occurs, why can't something be sent out at least 15-30 minutes prior to whenever the host decides to reboot itself as part of this self-healing process? At least give us a chance to notify people on our side so we can log out gracefully, etc. This is what we would do with a physical server under our control, at least.