Fork me on GitHub

moin moin morning


morning 🙂


ahoy hoy


Quick "straw poll"... How are you guys all doing the scripting of instances / containers (either is fine) that need to be provisioned, deps installed, code checked out / installed from S3, process run, results persisted and then instance or container killed? 1. Ansible 2. Chef 3. Puppet 4. Mach/Roll 5. Fugue 6. (ba)sh scripts 7. Elastic Beanstalk + scripts 8. Something else...


I need to figure out an interim solution that will do the above, for an R App that is being developed by my Data Dept - I prefer not to use Docker / Containers right now__ as I don't want the cost of an "always on" K8s / DC-OS cluster (even a small one) when I can do the above, for now, with an instance and cost myself a few $s a month.


But the more automated it can be the better I will feel 🙂


I have used 1 and 2 in the past and prefer 1.


hope that helps


@thomas - Yes, thanks, that does help 🙂


I also like Ansible, though I've never personally "used it in anger" it's been used at two of my previous jobs and I had to understand it and I liked the way it works... I find / found Chef "bumpy" by comparison, but I know that some people love it, so I am not agin it, per se, just not wild about it either...


I am interested in whether or not anyone has figured out a "good" way to do this hosting the recipes / playbooks etc. remotely to have the automation(s) run on a schedule without the need for a human being to "push the button" as well...


Side Question - is it worth developing my own custom AMI for something like this, or is it much of a muchnes to just start from baseline (say Amazon Linux) and script installing all of the deps on top before adding / deploying the application that is to run every time?


@maleghast i'm just moving to terraform for provisioning clusters... mostly because there exist some terraform repos which do most of what i need


@mccraigmccraig - Yeah, I like Terraform... How do you automate its use though? I mean I like what and how it does what it does, but it's not a complete automation solution, right?


Is it complete enough that non-fragile shell scripts can do the activation / running part of the process?


i think you would need something else too... possibly just a bash script


I understand that Ansible plays nicely with it, for example, so you can have a very simple Ansible playbook that can be automated easily in terms of running it / when it runs, but it hands off all the provisioning to TF


*nods* Yeah, cool 🙂 Simple is best. I need to refresh my bash skills, though, or make a decision to use "something else". I've always liked Python for that kind of work... I suppose one could use Clojure and Boot


for our CI testing automation we are using codefresh, which is basically docker compositions as a service - which works very nicely for setting up a clean test environment, running tests and tearing down


i think there are a bunch of other docker-composition services too


Hmmmm... That might actually be an interesting "third way"


brb lunch...


Lunch good - I am managing to be "healthy" today, properly... It's an odd experience... 😉