Where's My Beer: Building a Better Kegerator with a Raspberry Pi & Sensu

Editor's Notes

• #3: One other note, how many of y’all know what a kegerator is?
• #5: Every superhero, er, project has a good origin story, right? Well, mine starts with glass bottles. Rather, my intense hatred of bottling beer. How many of you have ever brewed or bottled beer? > Well, you know my pain then. > For those that haven’t, bottling beer is a rather labor-intense process that even if you have new bottles, still involves sanitizing them and prepping them One of the other problems that I had with bottling is that every now and then, I’d have batches that ended up like the gif above. Every time I had that happen, a little part of my beer-loving soul died. So I made the decision to start kegging.
• #6: The thing about kegging is that you have to have something to store/dispense what you have in your keg. That’s why I decided on a BYOK, or a build-your-own kegerator.
• #8: So when I started making my kegerator, I quickly encountered my first problem post-production. I had a frozen keg. Basically, I had a delightful barrel-aged scotch ale that was nothing more than the world’s biggest alcoholic push pop...except it wasn’t going anywhere, and I had to thaw it out. If anyone of you decide to build a kegerator, I’d highly advise getting something that allows for a bit more control that your standard dorm room minifridge.
• #9: Ah, the second problem. Leaving the door open. This wasn’t nearly as a big of a problem as you might think (though my wife might disagree). For the non-homebrewers in the room, temperature is super important when it comes to the lifecycle of a beer..Some beers ferment at different temperatures. A lager ferments at a different temperature than an ale. When it comes to storing beer, temperature plays no less of an important role. But in this case, I wanted to make sure that my kegerator door wasn’t open, resulting in a fridge acting as an AC unit.
• #10: Now, the biggest problem (and admittedly one that I’m still in the process of solving) is that I’ve had when I’m consuming my homebrew is that when I reach the end of a keg, it’s only when I start getting a ton of foam in my glass that I realize, “Oh...I’m out.” This problem is SUPER embarrassing if you have company over, and they’re expecting tasty, delicious homebrew. I’ll get to how I’m solving this in a few slides.
• #11: Talk about missteps here: Other solutions Trying to roll my own
• #14: Here’s a list of the parts that were used to make the system that monitors the kegerator. I have a full build list in a repo that I’ll have up after the presentation. You can head there to get the list of everything I used and links to the sites where you can buy them.
• #15: NO REDNECK
• #18: Who has ever used Sensu?
• #19: Sensu is a monitoring framework We make monitoring software that makes it so you don’t have to go out and build something yourself from scratch. Sensu has two products: Core Open core A very active community that write and hack on the product Enterprise Has support Come with some super slick interations that make building out a deployment that much easier
• #20: Besides the fact that I work for Sensu, why did I choose to use it to monitor my kegerator vs something else, or building my own? I started off wanting to roll my own solution. While great, this was intimidating for someone who doesn’t code a lot. I began to think of What would the alerting do How would it function What about notifications like Slack, email, SMS, etc. I quickly realized that rolling monitoring, and something like a dashboard, would be quite the project and decided that since Sensu already does monitoring well, and gives me the ability to decide on pretty much anything when it comes to a monitoring pipeline, that anything I did would be reinventing the wheel, and not in a necessarily good way. Why else? Sensu’s plugins and extensions are easy to write. Keep in mind that this is coming from someone who’s never written any Ruby that’s been put into production. Sensu’s libraries are suuuuper easy to use and super easy to understand. Oh, and did I mention that you can write plugins in any language? Besides Ruby, you can write in Python, Powershell (I know, probably no Windows users here at a Linuxfest ;) ), or really any language that is going to provide an exit status code that adheres to the Nagios api. Now, here’s the other thing about Sensu: I can run it on pretty much any system (MacOS, Solaris, AIX, and Windows included). But in this case, I can run it on a Raspberry Pi. Now, there’s a bit of finagling that has to happen to get Sensu to work on a Raspberry Pi properly, but I’ll run you through how I got it running.
• #21: So I’ll go through a brief overview of how Sensu works. This diagram is a simplified explanation of how all of the components of a Sensu deployment work. The hardest thing for most folks to wrap their minds around is the transport bit. This is the crux of any deployment. Without it, the server and the client can’t talk to each other. As with most other monitoring systems, there are also server and client components We also have a datastore for storing event state data, an api to programmatically interact with the datastore and server There’s also a dashboard (called Uchiwa) that allows you to have a more graphical view of what’s going on inside of your Sensu deployment
• #22: Here’s the general gist of how everything flows when it comes to Sensu. For those that were in the workshop yesterday, this should make some sense. For those that weren’t, I’ll walk through (briefly) what happens. With Sensu 1.x, our core components for any installation are what you see in the middle on the server there: RabbitMQ Redis sensu-server/api Uchiwa So we take our raspberry pi, with some minimal Sensu configuration on it, point it at the
• #23: Earlier, I’d mentioned that Sensu has configurable pipeline. This illustrates what I mean. What you see here are the fundamental building blocks for Sensu. Don’t worry if these don’t make sense right now. You’ll see how these work together in a few minutes. Suffice it to say that each of these elements has a role to play: Checks gather data and create events Filters only let important or relevant events come through A mutator changes data (if I need it to) And handlers tell somebody, or something to take action
• #24: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #25: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #26: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #27: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #28: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #29: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #30: So Sensu has this concept of a plugin. Plugins are these things that allow anyone to add functionality to Sensu. In this case, the plugin that I’ve written add the ability for Sensu to monitor any sensors that we attach to the Raspberry Pi.
• #31: At this point, it’s time to show how all of this works. So what I’ll do now is flip over to my Sensu instance and show off some of the magic.
• #33: Some folks have asked how the raspberry pi communicates with Sensu Explain