Why is the trend today towards installations like this?
curl install.amon.cx | bash
I'm looking at you, rvm, npm et al. Yeah, I'd just love you to poop files randomly into my filesystem, with no tracking, and a completely non-standard uninstall process (if I'm lucky enough to have one at all).
We evolved package management for a reason - I know it's a pain to have to make a .deb, a .rpm and a .tar.gz - but is it really that much more work than making an entirely custom install script?
This, along with the dependencies (I don't like having to install MongoDB just for this), is the reason I don't want to try this.
I could probably deal with the installation, although I hate installing all those daemons and init scripts myself, but installing Mongo just for one thing, gah. Too bad there isn't an embedded version of Mongo, and my own goatfish[1] is still far from mature.
It looks like a great project. Some features I would like to see:
* Alerts via SMS (possibly by integrating Twilio / Tropo).
* API via zeromq instead of HTTP. Async logging via workers. This could become handy if you log lots of things. It would also prevent logging to become the bottleneck of my applications.
This seems really interesting and I really like that the data would stay inside my firewall. Any comparisons to New Relic? Looks like ease of installation would be similar (except for setting up Mongo), but that it doesn't collect quite the same amount of data (yet).
Amon appears to have a slightly different goal set than Monit. Monit is more of a "Is this process running and ok?" tool for keeping processes running as expected. Amon seems a bit more like Munin.
Monit is a bit more than that. You can also use Monit to check remote hosts. Not only can it check http, but ftp, ssh smtp and more. You can use Monit to test your web-application via the HTTP protocol (404 or 50x) and use regex to check if certain content exist (or not) on a web-page. Monit is free, open source and does alerts for free.
Thanks for the clarification, but my intent wasn't to describe the full feature set of Monit. All those things you described are still limited to the scope of "is X running?" type questions. Can Monit graph time series data?
EDIT: My point isn't that Amon is "better" than Monit, only that it's different. There is some overlap in feature set, but ultimately, the tools have different goals.
It looks like it isn't possible to add multiple user accounts (in the free version). I tried browsing to /create_user myself to add a second account, but I get redirected to the /login page
Is that something that comes with the Plus version ?
That's something I've been working on for some time, but it's not ready for public release. As I see it - Amon Plus will have a fully featured ACL in a week.
Besides the fuzzy feeling of helping a open source project with comercial endeavors, how does it compare to services like Server Density, Nagios, Monit, etc?
I'm looking at replacing Munin with Amon. We used to use a pay-per-month monitoring service, but I find the setup and maintenance of many services to be equal to the headache of maintaining something like Munin.
The most obvious difference is that Amon runs on your hardware, so the data stays with you. The pricing model is fundamentally different than services like Server Density, New Relic, Scout, etc in that you pay per major version, rather than per-month.
Amon is currently pre-1.0, so it's hard to compare with a mature service like Server Density. They're (Amon) no where near feature parity with SD. Then again, SD will quickly cost you many times the licensing cost of Amon when you're monitoring multiple servers.
For me, the value judgement will have to wait. With services like Server Density you are (theoretically) offloading a lot of maintenance overhead to the SaaS provider. If Amon delivers on their "one line installer" promise, then it will at least have "effort parity" with many SaaS providers.
I have to wait because they're pre-1.0, so while their software is inexpensive and easy to use, there is a distinct lack of critical features. If I'm making the evaluation (whether to roll this in to production) today, I can't live without alerts, so Amon is out. Based on the way things are looking, I'd say that Amon has a bright future though. I might buy an early release, just to get up to speed on it.
Tangent: Where did you see that Amon is open source? I see that I can get access to the source by licensing it, but I don't see any mention of open source on the pages I viewed.
Also, on the open source thing, it's closer to open-core. The single-server version is open, but, if you want to monitor more than one box, you'll need the Plus version, which, despite being distributed with source, is not, AFAIK, open.
As this is self hosted, it leaves you to deal with all the management of things like alert delivery (e.g. ensuring delivery of e-mails, working with SMS gateways) and data retention.
The MongoDB installation that Amon sets up is not in a replica set nor sharded, so it leaves you to deal with redundancy, failover and scaling. We use MongoDB for Server Density and I can tell you that managing huge quantities of data is not easy.
Just wanted to let you know that in Internet Explorer 9 much of your page text is cut off. It seems to be positioned to the left of the browser so much of the text (mostly the code) is cut out.
The normal version can monitor only the server it's installed on. The Plus version has alerts and you can use it to monitor as many servers and apps as you like - there are no internal restrictions besides the memory of the VPS it's installed on. So if you want to monitor 10-15 servers - you will manage with a 128-256MB VPS, if you have more than that, all you have to do is to upgrade it to 512MB RAM or more.
So, is that something where the "central"/plus server polls the other servers that are running the normal Amon logger/monitor for their info then just displays it in one place?
curl install.amon.cx | bash
I'm looking at you, rvm, npm et al. Yeah, I'd just love you to poop files randomly into my filesystem, with no tracking, and a completely non-standard uninstall process (if I'm lucky enough to have one at all).
We evolved package management for a reason - I know it's a pain to have to make a .deb, a .rpm and a .tar.gz - but is it really that much more work than making an entirely custom install script?