How can I put a list of netdata monitored servers in the dashboard? For now I do not want to use Netdata Cloud, just multiple netdata instances (one database on each machine), all behind a nginx proxy.
The most obvious places well be a dropdown on the server name, or even better on the “chevron” that opens netdata cloud sign-in widget. But I could not find any documentation about the subject. It doesn’t need to be automatically (I will be very happy with some configuration/files generated by ansible or friends). Just how to “hack” the dashboard to include my list of servers.
I’ve tried in the past with the registry component, but with limited satisfaction (IIRC you must visit the node before it appears on list), and now the recommended way is to use netdata cloud in order to organize the servers, which, for the moment, I would not like to use (as personal opinion, already tested it)
You can always create custom dashboards and put whatever you want there, but it sounds like you want something done automatically, and I’m not sure what that could possibly be. One suggestion I can provide is to use streaming/replication and connect all your nodes to a single parent. Then you’ll be able to access each node’s charts via the parent’s UI, without directly visiting any children.
Thank you for your answer!
Very interesting idea with custom dashboards, probably it’s exactly what I want (to “hack” somehow the default dashboard layout and do some modifications). I’ve already read some docs, but I thought there was some “official” way to do small modification on layout (i.e. some javascript arrays, parameters, files on disk etc.).
From the netdata point of view it’s all manual, probably the server list will be generated outside (i.e. ansible template) and included somehow in the dashboard (but first I must understand the html code :-P).
It was a very tempting idea to use the builtin registry, but if I understood correctly only the visited nodes (and worse, separately for each person accessing netdata charts) will appear on the list, so it’s not the proper solution.
I had some test with netdata parent/children topography, but with 90%+ garbage traffic (unused data) it’s not interesting. Most probably I will use netdata for “live” (1 sec interval) data access and some other centralized data storage (?prometheus) for 5/10/x seconds interval data and longer “storage”.
Of course, netdata cloud it’s “sexier” than any solution, but for now it’s out of discussion, and afaik there is no plan for an “on premise” variant.
You don’t need to send all the data to a parent, it’s highly configurable. Check stream.conf itself, it has the necessary info.
For the sake of your sanity, I would advise against trying to reverse engineer the default dashboard. Many brave souls have taken that dive, never to come up again. A custom dashboard is a clean slate and it’s a file you fully control.