Are people actually running on a RaspberryPi?

Looks like with @garrettwp and @RichardTSchaefer making the move to openHAB, that I will now start to look into it seriously. I’m gonna read back through the thread, but what I’d want is to have it run under a VM (I have a windows 2008 server at the house for my media storage) or a very low power computer. I’m curious, since this board was started talking about it, if a rasperry pi would work. I want something though that will be powerful enough as well.

My only concern right now is that I have so many devices for my HA (Vera, ISY, Hue Bridge) and that adding one more just seems so daunting. My utility closet is already a sweat box as it is…LOL (granted, in the winter, that’s actually a plus :smiley: )

By all accounts, the RaspberryPi is incredibly slow to startup (read: minutes) and has a propensity to run in to OOMemory issues.

It’s only 512M, single 700mhz core, so it’s going to have limits.

That’s why I was looking at at C1, it’s the same price, and size. Still got a few kinks to work out, so the VM option is a good route to understand your real need before making the final decision.

I’ll chime in here on the hardware front. There are a LOT of different platforms from the OH community. You will find the OH community is extremely active and helpful, potentially as much if not more than the MCV community.

I’ve been through a few iterations of platforms myself as well. I run a simple Vera Lite and have for the most part standardized my HA around zwave. This leaves me with a simple box that has sometimes been overrun due to the quantity of devices and the Lite instability problems when it gets loaded. The basic premise for being able to run OpenHAB as far as I know, is you can install the Java SDK and place the OH core files and server components on the system. The overhead on the system is low, but as mentioned without horsepower it can be a bit slow on start-up and might run into memory overrun issues. The reasoning why, is because of how extremely extensible this platform is. So if you want to run on RPi you could, but performance may not be there. VM is perfectly supported, and I’d suggest a Linux OS for stability.

Now from a successful run category, I’ve managed to pick some of the obscure platforms. :slight_smile: I started this on my FreeNAS box initially. It didn’t run VM’s per se, but a jail which is similar. The base OS was FreeBSD vs the common Linux or Windows user variants. Mac is also supported, but if you have Mac hardware its better spent doing other things in my opinion. (I’m a Mac fan if I didn’t just give that away in that statement) I later decided to move my NAS platform to Synology. There is a package steady out there and required essentially just getting Java installed and the app content transferred over. Both had some learning associated, but again the community is extremely helpful in assisting in troubleshooting etc. @guessed has basically paved the way for easy mios integration and I would highly suggest digging in. Since switching, I’ve never been happier. This system is leaps and bounds more extensible than Vera by itself and easily brings multiple systems together into a single interface. I was able to easily make some of the rules I used to rely on @RichardT plugin for. The system is extremely lightweight as well and makes mobile access very easy as well.

I’m sure I don’t need to sell OH, so to finalize on the platform discussion, test on your local desktop in a VM and I’m sure you’ll be up and running in no time. If you have a virtual platform of choice you can likely run on that server as well afterward. If it’s mios related issue post here (were focused obviously) and other general system issues post or search on the Google Groups. People are quick to respond and help out.

A photo for sizing reference.

Top Left: Vera3
Bottom Left: Wink

Top Right: Odroid C1
Middle Right: RaspberryPi B+, then a Aeon USB Z-Stick 2
Bottom Right: Ubiquiti EdgeRouter Lite (for @SirMeili)

I couldn’t find my Vera Lite, so I put coins atop the ERL for an alternative sizing reference (US Quarter, Euro 2, AU 20c)

The C1 will be migrated to my wiring closet shortly, for now it dangles unceremoniously from the wall.

NOTE: The Odroid C1 has been out ~1mo. The more commonly used device is the Odroid U3 which is larger, more expensive and has a heatsink… but is more stable because it’s been out a longer time.

@guessed - uh oh now you’ve opened pandoras box! I see you have the Wink Hub there. Have you rooted it according to the recent circulating information about rooting them? If so, does it allow you to manually add or enroll ZWave devices that were not originally intended for the Wink Hub? And have you tested it out with the early OpenHAB binding capabilities I saw detailed in the Groups board I saw? I’m extremely interested in this and might decide to invest in one if I hear of enough success with OpenHAB. I’d like to use the GE Bulbs as they are cheap and awesome, but I can’t get them into Vera or OpenHAB without the Phillips bridge. I’m tired of buying extra hardware only to create another plugin method for another technology. Same thing with the WeMo stuff. So if we can get a Wink box working efficiently, I’m all about jumping on getting one.

At the moment, it’s just waiting in the wings. I’ve avoided firing it up, or patching it, so it can be rooted when I’m ready (and, if that’s still needed at that time). [IIUC] You need to root the Wink, and put on a fix, to use any form of API without being forced to use their cloud service.

The Wink, and the Connect, were free and/or cheap (at the time) if I bought certain other items. They’re just play-toys for when I need to integrate them directly with openHAB.

@Guessed - you’ve made my day: “US Quarter, Euro 2, AU 20c”. It’s good to see someone actually realises/realizes there is more than one country in the world. And that many of them - erhhh pretty much all of them, don’t use Fahrenheit ;D

I’ve tried out OH using a Windows 7 box (yes, yes; my apologies) and it works fine, once you set the PATH to the Java stuff.

A couple of people are working on a Vera MQTT plugin. I’m not sure how that is progressing? Guessed - do you have any views on that approach versus your mios interface binding? While MQTT could be labelled as the flavor of the month, it’s getting a lot of support eg MySensors and these associated apps look very interesting:

mqttwarn - for notifications
Own Tracks - for geofencing

On the later, if any one knows how to obtain and set up certificates for a Window box to make “Own Tracks” secure, please let me know.

More single board computer info:
More detailed board comparisons 1
More detailed board comparisons 2

I am really appreciating this thread and the great work that @guessed has committed himself too. Thanks for the great work. Enjoy the trip to Vegas.

When I finally get home at the end of the month, I am looking forward to setting the wheel in motion and this is my 20c AU.

Glad you like it, it was kinda tongue-in-cheek 8)

Next time I’ll use more colour…

A couple of people are working on a Vera MQTT plugin. I'm not sure how that is progressing? Guessed - do you have any views on that approach versus your mios interface binding?

Vera is generally strapped for resources, and that’s not going to be fixed anytime soon. Even all 3 of the new models are capped at 128M, at least according to their flyer @CES 2015.

The primary goal of the MiOS Bridge Binding is to enable folks with larger systems/deployments to expand out to something bigger that could handle their real automation needs, move beyond basic control, and still have a stable [overall] solution.

I took Vera’s resource-limit as a key driver, and that’s what pushed me to use the MiOS Remote-Control interfaces (the path-well-trodden).

The theory was that this would give me overall stability, a high level of code testing/coverage, and overall support/performance - all without having to ask the MiOS team to lift a finger (and, as we can see, they haven’t)

Anyhow, the Remote-Control interfaces have been used for some time, and are well tuned to “bulk” data pulls from Vera, along with the associated models for incremental pulls over-time so, for me, they fit the bill [almost] perfectly.

That said, they do have their issues but, for the most part, they work really well (just don’t be too reliant on Vera at the time of the 2:00am HEAL)… the birds are chirping on that bug, and the others uncovered during the process.

Any other option would have required installing something onto Vera, further increasing it’s performance/memory burden, the install complexity, and the likelihood of it falling over more than she currently does. So I dismissed those, for now, but they can always come back if there’s a compelling reason for them (like, say, Vera itself using them natively)

While MQTT could be labelled as the flavor of the month, it's getting a lot of support eg MySensors and these associated apps look very interesting:

mqttwarn - for notifications
Own Tracks - for geofencing

Sure, why not JSON-over-HTTP, or CoAP for that matter. There will always be a new-new thing. CoAP and MQTT are both players in this [embedded, lightweight protocol] space and an argument can be made for either (they both have strengths/weaknesses)

… anyone remember CORBA, SOAP, XML-RPC, the list goes on…

In my case, why add an extra layer in between openHAB and Vera?

When I want MQTT, or CoAP, or some other protocol x-change format I’ll add it where there’s resources - notably to the openHAB end where users can choose to install the option if they want (or need) it, where there’s processing power to handle it (and any newcomers that rock-up down the line).

I am really appreciating this thread and the great work that @guessed has committed himself too. Thanks for the great work. Enjoy the trip to Vegas.

When I finally get home at the end of the month, I am looking forward to setting the wheel in motion and this is my 20c AU.

It was quick and a lot of fun (yes, yes, that’s what she said). There’s a full day of just walking the floors @CES with a lot to see in both CES-East and CES-West.

eg. Nortek had a nice [ZWave] Water-flow detector that doesn’t require the pipe to be cut. Will be handy for leak detection, when correlated against typical usage patterns (or other historical data)

I also got to talk @strangely’s ear off, and he mine… I’m sure we’re all set for another year now :wink: