Hacker News new | comments | show | ask | jobs | submit login
Huginn – Build agents that monitor and act on your behalf (github.com)
264 points by metachris on Feb 1, 2016 | hide | past | web | favorite | 44 comments




And the biggest ( opensource) alternative is :

- Bipio @ https://github.com/bipio-server/bipio


Then you have to deal with both nodejs and mongo though.


"Ah Huggin ... not again .... " - Every-time I see it on HN.


A little surprised nobody has noted the Daniel Suarez connection as yet. This is obviously inspired by "Kill Decision" (note the logo of Huginn wearing a headset) and "Daemon".


So there are hidden gems in the comments(at-least for me, I did not know about Node-Red, Bipio etc). Thanks for sharing!

Is/are there any such framework/s built with/in Python too?


Check Ramses [1][2] + Elastalert [3]. I'll be happy to help.

[1] disclosure: I'm one of the core dev [2] http://ramses.readthedocs.org [3] https://elastalert.readthedocs.org


Dude, this is what I've been building for a while now. this is awesome! Thanks!


I've recently started a project to write an automation server in Python. It's still early days, so I'd be interested to hear what features you'd like to see? I'm aiming for easy extend-ability at moment - drop a module into a directory, reference it by name in the configuration file, done.


I would love to use and contribute to any such project in Python or Java.

Feature wise, I would say should start with everything we have in Huginn.


> Is/are there any such framework/s built with/in Python too?

What an awkward regex! :)


:-) Feel free to suggest! I hope it conveyed the idea. Is there any such framework built in Python too? Are there any such frameworks built with Python too?


I am looking for a Python implementation as well.


I really like this concept -- in a previous life at a large company, this was a real problem -- managing internal systems on a prehistoric RHEL version often required employees to build these faceless users who were authorized to do stuff, but I always thought it might be better to just have people run something like this -- a personal agent, that software would trigger/alert.

Glad to see someone creating a manager for this kind of thing.


I actually used this for a good while and I like when Huginn shows up. Andrew Cantino is a great dev and has put a lot of hours into this project. As well as many other contributors.


Thanks! But lots of other people have worked really hard on Huginn as well.


That's why I wrote and many others as well, I just don't know who anymore. I feel bad that I've lost touch, perhaps it's time to submit some more pull requests.


OT: I learned about Odin's two Ravens two days ago. Baader-Meinhoff is in full effect.


Hi,

I'm a software engineer at a small startup that makes this open source engine that automates environments orchestration using containers.

http://azk.io

We created a super easy way to quickly (and safely) run applications such as Huginn from source code locally on your own workstation.

It takes little more than just a click.

Go to this fork of the project and click the "Run Project" button:

https://github.com/run-project/huginn#running-locally

The only addition made to the original project is a manifest file about the OS, languages, databases etc. required by the project.

Afterwards, if you want to deploy to DigitalOcean from your desktop, you can do it in very few steps following these instructions:

https://github.com/run-project/huginn#deploying-to-digitaloc...

It would be great to get some feedback about it. Hope it helps!

Thanks.


Or, if you wanted to skip the self-deploy and upkeep of Huginn, there's now a hosted instance at https://Huginn.omniscope.io/ with a sixty day free trial for those who want to try it out first reply


That's neat. It reminds me of Bitnami (http://bitnami.com/).

When we created azk, we had developers in mind.

Solutions such as those or the `Deploy with Heroku` button are cool, but they don't give developers the opportunity to just fiddle with the code or even just check how the application looks like running locally before deployment.

So we tried to expand on those solutions by adding this capability on it (and you can always easily deploy to DigitalOcean with azk).


This is a great service, but I couldn't configure a Twitter stream agent because the installation doesn't have Twitter set up.

Another thing I'd like to know when signing up to a service like this is if and how I can get my data with me if I decide to migrate to a self-hosted version for example.


Seems interesting, do you mind enumerating what "Semi-private Workers" means?


Awesome!


I'm not really clear on what azk actually does...

Is this an alternative to Vagrant?

Why wouldn't I use Docker instead of azk?


azk is a higher level environment orchestrator than both Vagrant and Docker Compose.

Instead of dealing with individual virtual machines one by one as you'd do with Vagrant, you'll just write a simple system-level manifest file (Azkfile) listing the environment's elements and the relationships between them and "hit play".

azk builds the environment for you with little intervention, in a short amount of time and you can easily replicate the results in other machines using that same Azkfile.

With azk, automation does NOT come at the cost of opacity. You can always look "under the hood" and see what's going on.

Today, azk employs Docker, but frankly, we could use any other equivalent solution (as we are currently testing).

Compared to Docker Compose, we can list the following advantages:

- Its manifest file (a DSL in Javascript) is more verbose and has a greater toolset [0];

- You can persist content from inside the container way better than simply mounting volumes using docker compose [1];

- For now, it relies on VirtualBox (usign debian2docker) but it has a built-in function [2] to sync folders into the container using rsync (significantly increasing performance for file sharing). You can still use VirtualBox Shared Folders if you want to;

- Its built-in DNS server makes you don't worry about port mapping and docker IP issues. You can access your local app running with azk by simply accessing a custom address, like: my-app.dev.azk.io.

We like to think of azk as the YouTube to Docker's Adobe Flash Video.

We've built a gallery [3] with some projects where you can have a first experience with azk by running them. If you get interest, you can learn how to write Azkfiles to your projects by accessing [0].

This three and half minutes video shows azk in use [4].

[0] http://docs.azk.io/en/azkfilejs

[1] http://docs.azk.io/en/reference/azkfilejs/mounts.html#persis...

[2] http://docs.azk.io/en/reference/azkfilejs/mounts.html#sync

[3] http://run.azk.io/

[4] https://www.youtube.com/watch?v=J2i8qug99Kw


This is an awesome project. If you want to do some lightweight tasks, you can get it running on a free openshift instance very easily.


Not the same thing. But similar = www.stackstorm.com or better https://github.com/StackStorm/st2 Disclosure - I'm a co-founder so arguably most biased person possible. Think IFTTT for Ops. Prior conversations here include: https://news.ycombinator.com/item?id=10342000


The concept seems nice and even like something that would be useful for me, but given my past experience with Ruby-based web apps, I have serious doubts my old Athlon XP server box could do a particularly good job of running it... is there any similar project written in a more performant language?


This would be a really fantastic program to have running in Go. Deployment is drop-dead simple and usage on resources is very light.


The actual computational work this app is going to be doing is rather small (waiting on IO). I would be more concerned about wasting memory leaving this running all the time.


I remember seeing Ruby on Rails constantly burning something like 10% CPU even when it was ostensibly only waiting on the machine (though this might be an issue that has been fixed since, I am not particularly optimistic given the day and age), and anyhow, even in the RAM department, I only have 512MB to work with (of which most is in use most of the time anyway).


"most is in use most of the time"

I don't know how you measured that, but Linux will use all of your memory all of the time with various caches.


Lot of interesting use cases for this. Very cool.


I fail to see how this is different than Node-Red. Node-red already has a nice ecosystem, and extra modules allow usage of any npmjs modules to be used (in excess of 200k).



And bip.io sits in between the two. I wonder why 2013 was the year for this sort of project.

https://github.com/bipio-server/bipio/graphs/contributors


Iirc Yahoo pipes was discontinued


I haven't heard of node-red so thanks for the mention. I just started using huginn for a month or so, and I really like it. I mainly use it to process input data from various sources and output RSS feeds. Looks like node-red doesn't have something pre-built for this even in the extra modules. It is on a list of ideas: https://github.com/node-red/node-red-nodes/wiki/Node-Ideas

So for now, if your usage is the same as mine. Perhaps huginn would work better.

I do like how node-red have that drag and drop interface though.


Is this what you're talking about? http://nodered.org

Seems interesting but the metaphors are rather different.


That's correct.

They are a bit different. However, with many of the node-red-contrib modules, I can do things like "watch twitter for certain #tags, send them to IRC along with a semantic analysis. Also log to MongoDB. And any @ send to my email every sunup."

The basic functionality of the core nodes is already impressive, but the contrib is where Node-Red shines.


huginn isn't tailored for IoT specifically.


"Great Odin's Raven!"


This might as well read "Scalp ALL the tickets!!!"




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: