"stop" or "pause" cluster

[dylanbannon]

It'd be good if we could just pause all the instances, thereby putting instance costs into abeyance, without tearing down the entire cluster.

[dylanbannon]

This is being addressed by the cluster-pause branch and pull request #63.

[dylanbannon]

There have been a series of pull requests (including #218, #247, and #217) since this issue was filed that, in the opinions of @MekWarrior, @willgraf, and myself, obviate this issue.

Our reasoning is that, when the issue was filed, cluster configuration was a hassle, so this enhancement would allow users to sidestep not only the waiting period for cluster destruction/recreation but also a convoluted re-configuration process. The pull requests mentioned above essentially removed the re-configuration step.

So, what are the advantages to a user for being able to choose to pause a cluster, instead of destroying/restarting? In our opinion, the only advantage is avoiding the cluster destruction/recreation time. Currently, this come out to ~20 minutes, so we're going to say that the inconvenience doesn't merit the effort necessary to create this feature.

Closing, until a more compelling justification for this feature comes to mind.