Hi All,
I’m working my way through a few features I haven’t touched before on LibreNMS, and at the moment I’m looking at the device dependencies feature to reduce alerting when a device such as a core switch goes down, (which would otherwise produce hundreds of individual alerts) and also hopefully also make use of the device dependency map to get a visual overview of the dependency hierarchy. *
There are about 320 devices monitored in total and I have started by defining the core switch where most of the servers including LibreNMS is connected as the root of the dependency tree. On this switch the children are the servers directly plugged into the switch and the three switches that are directly uplinked from this switch. So far I have not gone any further than this.
However I’m having a lot of trouble trying to get any sense from the graphical Device Dependency map.
The main issue seems to be that all devices are included on the map even if they don’t have any parent or child dependencies defined. this causes a huge unmanageable “cloud” of points like this:
Is there any config.php setting which can hide devices which don’t have any device dependencies defined? I’m not interested in knowing where every single device like every printer etc is, I’m really just interested in mapping out the dependencies of the main network backbone - eg switches and servers in the dependency map.
Possibly related to there being “too many” points on the graph, even the dependencies I’ve defined so far don’t seem to display correctly. If I zoom right in to the switch that is defined as the root parent I only see a couple of the 16 child devices, and the graph is jumping around like a rubber band:
About 12 devices seem to be stacked on top of each other at the bottom, and if I pick at them with the mouse and try to drag them apart I can separate some of them but not all of them.
I’m not that familiar with the underlying graphing system but it seems that there is too much “pressure” from the surrounding point cloud of other devices without dependencies to let the graph expand properly, hence the desire to hide devices without defined dependencies.
Anyone else use the Device Dependency map successfully and/or found a way to hide devices without dependencies?
*Yes I know about the automatic network map (generated from LLDP etc) and I do use that as well, however quite a few of our switches do not support LLDP so the map is incomplete and misleading in places (especially where two or more LLDP capable switches are separated by a non-LLDP switch, this shows weird loops on the map) and it also doesn’t show servers, since they also don’t run LLDP.
So being able to manually define a network map topology of the core switch network and where critical servers are located using the dependency feature, and also have it filter alerts in a useful way is quite appealing.