Austin DevOps Meetup – InfoChimps present Iron Fan

Posted by gminks in meetups | Tagged , , , , , | Leave a comment

I hate playing favorites, but I really think the Austin tech meetup I enjoy the most is the Austin DevOps meetup. It is super technical. The presenters are very real – no agenda to sell stuff. They just like explaining how they are doing things. So far no one has seemed to mind if a huge discussion of “well you can do it this way too…” breaks out, along with all the typical scenario-building/busting banter that techies love so much. 🙂

I really learn a lot from these meetups. If you are starting out with cloud / devops tools and you are in Austin, I’d really encourage you to attend. It’s a great way to bridge the knowledge and experience you already have to the new tools that are coming out.

With that said – at this month’s DevOps Meetup, InfoChimps presented their orchestration tool called Iron Fan. My raw notes follow, also check out the blog post from Copper Egg (the hosts of the DevOps meetup).

Here are my notes – as usual please let me know if I got something wrong!

Iron Fan = orchestration layer.

cluster def = core of it dsl on ruby

group of machines that are coordinated together to do what you are describing

can copy a chef nomenclature

lots of stuff written for amazon

define cloud (instance, type. zone, machine, etc)

can define vols based on atributes, specific to hadoop deployment

defines the chef environment to use

ops – manages complexity of cookbooks, makes sure it stands up in production 🙂

scope levels:

cluster,

role, scope, facet,

role = chef role

creates two types of role on launch, cluster and facet.

broker objects = computer/disk (if present)

providers = gives you resources

iron fan is a knife plug in

fog version won’t be picked up by a long standing recipe.

uhoh devops fight!

person 1. why would you say you may lose a machine that’s not really devops

person 2. why would you say that?

image building is not done by iron fan.

cookbooks

silverware cookbook – announce and discovery mechanism – simplifies/standardizes chef lookup for X

“amenities”

announcements – one node can announce itself to others, nodes can connect (via running cookbooks)

chef server is persitence layer by default

hdfs backed (can be) by volumes

 

Leave a Reply

Your email address will not be published. Required fields are marked *