Deploy script does not handle errors reasonably

[petiepooo]

Connection errors (eg. because we must use a proxy) are ignored, and configuration continues as if the package was successfully downloaded. Rerunning produces many other errors due to partial configuration being present.
Scripts like the deploy script should really be idempotent.
It would also be nice to allow certain portions to be skipped, like if ntpd is allready setup, then don't deploy chrony, or if iptables is already active, don't deploy firewalld.

[peasead]

Thanks for the report, @petiepooo

I think the proxy issue is a good one, what would you suggest to fix that if the person deploying isn't using a transparent proxy?

I like the skipping idea as well. That said, generally speaking, CAPES is supposed to be deployed on a fresh build vs. a box with other services, like iptables running. However, golden images could come with those things, so point taken.

Would you be interested in submitting a PR for these issues that fit your use-case and we can see where to shim them into the project?

[petiepooo]

I am actually working on some changes now. I like the way you've gathered these tools together, so I'm working on the changes that work for me.
Npm is a bear to get working via proxy, especially when it launches git behind the scenes. I think I just got that working tho as I got hackmd running now. Expect a PR in a day or two. A bunch of little commits, some of which could be skipped if desired...