Wow, this is really cool. Sometimes I just keep tweaking a server losing track of what I changed where (that's just me), would be so easy to just convert all the configurations I did automatically into a puppet script and deploy it on a bare-bones machine!
So, how would they go about monetizing this thing?
Hey I'm Matt. Honest answer at this point - we don't know, hence the open-sourcing. I think our initial attempt last year was not explained well and too difficult to get started with. This is our attempt to simplify our initial work as much as possible. Blueprint figures out what you did to a server and sandbox uses chroot magic to give you an isolated container to work in.
Our current direction which should be open for testing in April is a hosted Puppet service. We're hoping to make it easier for people to get started w/ a devops backend (Puppet, Nagios, etc).
If you have any questions, comments or suggestions I'm all ears matt AT devstructure dot com
If this evolved into something like ninite for servers I would pay for that. I mean, install and configure apps that I select, secure the server, etc., also give me a file with all the passwords, configs, etc. that the script made so I can understand what it did.
I'm a developer, not a sysadmin by any means and it's daunting for me to setup a new server and then remember what and how I have re-configured. Usually I just follow my host's tutorials and never come back to tweaking anything.
Maybe something like that exists already, I've no idea.
This is cool. As a solo cofounder/CTO/developer/sys-admin this is what I need, both to evaluate what I can scrap from my boxes and in case of everything going south to rebuild stuff. Pitty my VPS runs CentOS now, assuming it's not supported? (only Ubuntu is mentioned)
We are actually going to migrate from one dedicated to another and documenting all the software, libraries, settings, and data has been a pain. Would love to try this but we don't run Ubuntu 10.04!
Side note: I am surprised that there is no floss migration tool that does this. I know that it would be a complex piece of software but I would actually pay/donate money for it.
Richard from DevStructure here. You're right about a lot of your criticism: the setup.sh process is not ideal. I do think it is preferable to pointing someone at packages.devstructure.com and saying, "Good luck!"
We'll be changing shortly to recommend `git clone` for first-time tinkerers and packages.devstructure.com for real use.
This is pretty killer. We use Chef, and although its repeatable results are nice, coding for chef and deploying via knife seems tougher than it needs to be.
This seems to have serious potential to position chef as a way of communicating configuration while requiring very little direct writing of code for Chef.
Definitely. We want people sharing their configs more, just like code. Secondly, we feel that for most people, configuring via the command-line is more natural, or at the very least an easier way to get started w/ puppet or chef.
Exactly this. Getting a config running is quite often an iterative and exploratory process that should not be enshrined in structure until it's actually running. You get this, which is why Blueprint makes sense.
Thank you for reminding me of this. I gave it a try a while back and thought it looked promising.
I recall having a couple of very small gripes with it (they really were quite minor and were likely due to my own misunderstanding), but I honestly can't even remember what those gripes were anymore.
Can't seem to edit my post but I remembered what my issue was. I was having trouble, at the time, getting it to work properly with RVM. I don't know if it was something I did wrong or if it was an actual issue. I will investigate tomorrow when I get back to work on the project.
I'm using Ubuntu 9.04 on my laptop and would love to use this to clone my setup. Is there a workaround or do I have to upgrade to 10 before trying? I've been putting off upgrading because I'm scared it will break something or the other as upgrades always do.
I am seriously excited about using this. I've been holding off on configuring new forward deployed servers before automation is in place, this lets me get work done now without worrying about shooting myself in the foot later.
I'd love to see something like this for FreeBSD (where a lot of the configuration stuff is already centralised to begin with), as it would make my life that much simpler!
Indeed, which I'm a big fan of. If we reach a point where it's confusing for people (and I'd be delighted by that), we'll consider a change. Since they're pretty different concepts I don't think it's a huge problem at the moment.
