Hi, We could do with exposing certain metrics from bgpd, ospfd and pf.
I was considering a couple of approaches and really was just interested in what would make most sense in general. Has anyone else considered this at all? Would this be useful to anyone else? Also just to say I am not necessarily suggesting this being in the core OpenBSd project, but happy to do that too potentially if it makes sense, just looking for some design ideas and experience, I am sure with each of the approaches suggested their may be things I just can't know at this stage, with my relative inexperience.... Option A) Create a daemon called ospf_metrics or similar which talks directly to ospfd and fetches and sames metrics to a txt file or similar (which the prometheus exporter i believe supports as standard) This could keep some data in memory such as rib counts or similar so it only occasionally has to query full lists etc. Option B) Create a daemon called ospf_metrics or similar which talks directly to ospfctl and fetches and saves metrics to a txt file or similar (which the prometheus exporter i believe supports as standard) This could keep some data in memory such as rib counts or similar so it only occasionally has to query full lists etc. Option C) Extend ospfctl so that it can output metrics a new flag, this could then be added to a cron or similar. I could use the new output capabilities in bgpctl and ospfctl to introduce a prometheus output (just worried this is too flavor of the month for a long term inclusion in ospfctl/bgpctl), and potentially the current cli options don't really necessarily match one to one with metrics you would want to emit. Option D) One of the above options but a new daemon called metrics or similar which provides metrics for various parts of openbsd. (similar I guess to snmpd) Option E) Extend the node exporter to do this natively in some way, haven't looked too hard at that yet, guessing could be a golang exe which talks to ospfctl or similar. Option F) A suggestion someone else has which is probably better and simpler... :) Cheers Richard
