I understand the reasons that Azure IAAS servers must be restarted, but the process by which it happens (for me at least) is really rough. Yesterday, All my servers restarted twice in one night, without warning of a planned maintenance window. From what I can gather this was due to host upgrades in the East US region. However, not only did it take each server much longer than normal to boot, the essential services (iis on web services, sql on SQL services, etc ) did not restart with out manual intervention. I have fail overs and HA, but it is of no use when i have to babysit all the servers so that they actually come back up before another server restarts. This does not happen when i restart the servers myself so why when MS does it?
In addition, our application is quite complex and has many layers and hence many availability groups - approximately one for each functional layer of our application. This means that the short outage that occurs while a failover occurs within an availability group, happens frequently during a host upgrade and so the application is unanavailable for a much longer period of time than expected during the short but frequent recoveries.
my questions is two parts
1. Has anyone else experienced services not restarting after an azure host upgrade and hopefully found a solution?
2. I was under the impression that availability groups should match for each functional set of services. i.e. one for the web front end, one for reporting, one for transactional DBs, one for load balancing, one for reporting, etc.. Is this incorrect?