Category5.TV Community Forum

Full Version: Distributed Monitoring
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2
I'm probably being a little ambitious here, but I know Nagios supports this config.

I have 7 clients with 12 physical sites between them, and my plan is to put an RPi3B+ on each site running NEMS. What I want to do is setup distributed monitoring, where all the traffic comes back to my main NEMS unit at the offce, and I can see everything on one big map and mount a 24" LCD at my desk and use the NEMS TV interface to watch for errors and warnings. My first 3 Pi's are almost here so I'll be doing some experimenting, but I thought I'd ask first.

Has anyone tried anything like this? Is there any documentation anywhere?

This is indeed highly ambitious, and not yet possible... though if you kinda peek over the changelogs and to do list, you might notice things that are driving toward this.

I'll be straight up honest and up front with you though: this is a big task. BUT know that it is something I am passionate about, and really want to see integrated into NEMS. This is one of the core reasons I moved NEMS away from Check_MK and onto Adagios... you'll notice if you spy on the config file, I already have comments there to integrate other NEMS servers via livestatus.

So... it's in the works. It's not the top priority [yet], and it's huge. A massive undertaking to implement.

Is it realistic, for now, to simply deploy NEMS servers at each site and remotely access them? You can open only the ports you want access to, and use a strong password. I took effort to ensure the private stuff is behind a password and SSL cert.

Alternatively, you could open port 22, setup key-based authentication, and use a SOCKS tunnel to access each NEMS server from off-site.

There are so many things to do before I can make this happen... I don't want to get hopes up thinking it's coming soon... I continue to work on it, but it's truly a massive feature!!

Another thought... I could do this in steps.

Like, I could setup NEMS TV Dashboard to support multiple NEMS servers... then work our way to other features.

If it could at least show which NEMS server had a device with an issue that would be a great start especially after linking it with Pushover.  My quick 2 cents but will also add more comments as I think about it more. Thanks!
Using the TV feature to monitor multiple NEMS servers would be a great place to start! 

Thanks for the reply!
mydogboris - YES, this is in the plans. The feature is being added to NEMS SST, so watch for it!!

readyit - Okay... I will put that on the immediate roadmap to get the ball rolling, thanks! It is still a big feature, and can be complicated (ie., users who aren't quite sure how to open ports in their firewalls or who have dynamic IPs could cause a lot of support issues), but I will do my best to make it relatively simple to setup.

The other option, readyit, would be for me to have the API hold 15 minute, anonymous data for each NEMS server... then your NEMS TV Dashboard could pull it from the API. Hmm, this could solve a LOT of issues for setup, and make it dead-simple for novice users. Only thing is, it would be very limited. But it'd let you have one display showing many NEMS servers. Thoughts?
Pulling the API data sounds good, but wouldn't that still be a firewall issue for those who aren't necessarily experienced in proper config'ing? Or would it be a push to the central unit?

In a pull case, or just a connection via firewall case, you'd almost always have to set a static IP on the NEMS unit - something I've done anyways because I can't trust that my DHCP server (Windows server on a VM) will come up to assign an address, which means I wouldn't get data from NEMS at the client site.
No that's the beauty of an API. I'd use a single identifier to unify your units (for example: login to your account, add your HWID's), and then use CURL POST to the API to talk to all the units, and a CURL GET to receive the information. The only static items has to be my API server... which obviously already is.

It's very "easy" for me to code such a system - just a fair bit of work. It won't be instant, but a LOT easier/faster than an actual connection between servers.

Probably a neat idea eh?

That actually sounds pretty slick.  I like the idea of that (even though this isnt my thread lol)
You can annex threads to give thumbs up :) After all, it's a community effort!
Pages: 1 2