Open hammer opened 8 years ago
There's actually a fourth, and even more different use: for constructing the Biokepi.Machine.t
I prefer keeping them together for this; there's overlap between them all (I think 2 is a proper subset of 3, for example): I could be missing the advantages of splitting them up.
There's some clarity improvements, certainly, if you're interested in what's going on under the hood; that could be cleared up through comments & more documentation, for the curious, but I don't think it's necessary for quickly getting a cluster up and running.
That and I've certainly failed to source configuration.env
enough times to dread having to remember to source 4 of them.
My biggest issue is that I'm trying to figure out what needs to be uncommented for each use case.
I guess I'd be okay separating the configuration out into sections with clear comments about what needs to be set for each use case.
Some configuration variables get reused between steps.
And it is used to hide form the user some annoying things to think about: like the SSH-CONFIG that gets generated locally and deployed on the Ketrew server, and then the pubkey is deployed on the PBS-cluster.
I see three places where it's used:
sh gcpketrew.sh
on GCloudernfs_server.ml
w/instratocumulus
Docker image on GCloudercluster.ml up
w/instratocumulus
Docker image on GClouderWould it make sense to have 3 separate configuration files, one for each use case?