Dear all,
Just wondering if there was an appetite out there for expanding the interface labelling feature of observium. The existing description parsing (http://www.observium.org/wiki/Interface_Description_Parsing) is pretty useful but also a bit limiting at times, as it seems more useful when labelling layer 3 services and less so for layer 2. Just a few general comments:
* For us the pseudowires tab does not seem to auto populate (we use a variety of juniper, brocade and extreme switches and routers), just wondering if I have missed something here...
* This aside there are more frequent occasions where we are providing VPLS services for a client and it would be nice if it were possible to label an interface as both a customer and a P2P link for example and then have these separately broken down in observium itself
* There are also occasions where it would extremely useful to label backhaul circuits such as dark fibre and waves. Currently the only option for this is to label things as core but that page becomes so big it is not that useful unless you know what you are after (and then you would just go straight to the interface). Something like BACKHAUL: , PROVIDER:, END (A/B etc.) would be immensely useful and time saving.
* The other issue we have that I don't have a simple answer for is the unhelpful aggregation of customer switch ports (quite often both ends of a P2P) and their layer 3 router ports. Could we start by having a separate aggregate graph for each circuit reference for each specific client. Maybe if END A and B are specified then the graph could demonstrate that these are actually different ends of the same link?
What do people think?
BW
rob