Chef redux

| | Comments (0) | TrackBacks (0)
Welp. Our visualizer product has been canceled. That's disappointing, to say the least.

I thought that meant most of my chef efforts and vagrant efforts were not going to need to go into production. But, it turns out, we still need builds, and we still need efficiency.
I keep finding myself creating one-off package builds, and hand-verifying things. And my co-workers keep sharing 'build out a workspace' scripts. So, it's like they're almost with me!

Feeling the need to make good on my goals of infra that stages and recovers. I can't (ought not!) write the recipes if I can't test them, and I can't test them until I have base box workflow.

The other day, I realized there's no reason not to slap chef client onto all my production and staging nodes. So I took care of that with knife, until I can normalize the rollout of new boxes from cfm. My next post will be about getting powershell and winrm on win7 so I can knife bootstrap things, and use winrm provisioning. Because that sucks hard in chef land.

When I rolled out my latest windows desktop, I looked at chocolatey. Super useful. Very convenient. There's a handful of things that I need to slap onto my machines, so a base image plus cfm-managed chocolatey rollout is very tempting to throw together, to share with my coworkers. I'll be visiting packer-windows soon.

I took the time to throw together vagrant osx base boxes for parallels and virtualbox (and, some team members use vmware, and we have some esx/esxi infra), so those tests can get written.

I'm left, while I slowly move in the right direction, wondering how users ensure they can bootstrap the whole pipeline. Eventually, something has to come back to pressing the 'get source and start the factory-creation' button. Chicken and egg?

0 TrackBacks

Listed below are links to blogs that reference this entry: Chef redux.

TrackBack URL for this entry:

Leave a comment

Recent Tweets