"Always-on" is deprecated (I believe that it is no more). The configurations that currently allow you to do this and more are documented at https://cloud.google.com/appengine/docs/java/modules/ ; for example, if you always want one instance to work, no more, no less, set the default module of your application to manual scaling, with instances set to 1 .
However, automatic scaling with min-idle-instances should also work if you followed the instructions (at the URL I provided)
In order for resident instances to function correctly, you must be sure that heating requests are turned on, and your application processes warm-up requests.
Use, as soon as this URL recommends, the "Availability" column on the "Instance Console Instance" page to check which of your instances are actually resident (as you wish) and not dynamic (i.e. leave after some time just me).
source share