When I update the version of a anw web app on a cluster of application servers I first install it on a single server, run a set of tests on that app and only if they all pass I continue to the next application server. The tests are usually quite simple and fast so I call then sanity tests or self-tests but they could be as elaborate as one wishes.
Ideally you'd want that exponentially to boost the deployment time, e.g. first start with one app server then two, then four.
Does glu support this kind of workflow? I haven't seen this on the wiki.
Are there any plans to?
* when using the UI, you can select 1 server only to be part of the plan. There are various options available as well, like 25%, 50%, etc... So you can do 1 first, then do more the second time, etc...
* when using the cli (not released yet but coming very soon), there will be a dsl to write your own deployment plan, thus allowing you to specify the kind of behavior you want
So in other words it is not supported out of the box but I think there are ways to do what you want to do (I really think that the cli will most likely be the way to go as it is complicated to do in the ui...).