We are having problems with the Azure Application Service. One of our web services (MVC) caches data from the database at startup (Application_Start) - it takes about 3 minutes. Until this is ready, we will not be able to process requests.
This is known, so we set it to "always on" and will strive only to restart it after hours, if necessary.
However, we expect a heavy load on the server next month, and during our testing of autoscaling we found that when it adds additional instances, each of these instances goes through the same startup delay, but the traffic is split between the current executable instance and the new one, which is warming up for example, half of the requests begin with a failure for this 3-minute period.
How can we configure Azure to delay using a new instance until it is ready? (or should we use, for example, AWS?).
Some documents point to the use of the custom Load Balancer Probe probe, but it mostly talks about VM, whereas we use PAAS.
Lance source
share