I’ve been playing with ansible a bit lately, and really enjoying the ease of the push-mode tool. I miss the strictly declarative nature of DSC, but it’s so nice to be able to just point a play (set of tasks, like a config) at a remote machine, and have ansible take care of copying all dependencies out there, running them, and returning all the results.
Best of both worlds would be great, so I thought what if you could have some sort of client on your dsc authoring machine that could compile your mof and deploy it to the target just like normal, but they also configure the LCM to use the authoring machine as the reporting and resource repository. The authoring client could then serve up any dependent resources, wait for the deployment, and receive a status via the reporting resource.
Of course for that to work your authoring agent would need to also be a web server, and you don’t just want to leave those lying around all over the place. But then I saw this project Polaris thing and thought hey what if you could build a little mini pull server in this project Polaris, and bundle it up with this authoring client thing. Then deploying a config would also involve starting up the pull server, doing the things, then shutting it back down.
Of course this all depends on what they do with dsc in PS6, and that is still TBD apparently? But curious what anyone thinks, if this is something that would be remotely useful or if it’s just a hack that would probably break more than it fixed.