Hi, See http://git.io/SlingDevOps-v4-demo for a demo of the work done by my intern Artyom Stetsenkos's on zero-downtime upgrades of clustered Sling instances.
This is still an early prototype where you have to start and stop the Sling instances manually, but it already demonstrates upgrading a cluster of Sling instances atomically, by just pushing an updated Crankstart definition file [1] to a Git repository. We hope to be able to demonstrate the full automated scenario soon, where you just push to Git to get your cluster upgraded. Note that this is fully experimental so far. I think the systems approach of zero-downtime upgrades, where the individual instances are immutable and throwaway, is worth a closer look at least for production instances where you want strict configuration control, ideally based on version control systems. Enjoy, and feedback is welcome! -Bertrand [1] https://github.com/ArtyomStetsenko/sling-devops-experiments/blob/master/crankstart/sling-minion.crank.txt