Closed avluis closed 1 year ago
Are you looking for a specific NTP MiB or just genetic system ones?
If rfc5907 {NTPv4} is considered alongside standard (system) OIDs I'd be a happy camper
NOTE: multiple people are now asking for this
I just ordered an NTP250 for my homelab. I have been perusing the documentation in preparation, and so far, I like everything about this appliance except that it seems to be missing SNMP and a legit serial port (which I believe might be a better place for a PPS signal than an unprotected circuit). SNMP would be a great addition that I would most definitely use to graph from Observium and alert thresholds from Nagios.
Thanks Troy, I started on this a few months ago and adding system MIBs isn't a big deal, however finding something for rfc5907 was a bit more troublesome. Threr's a PoC AgentX that implements parts of it out there from a GSOC project but it's not really useful as is. Implementing this MIB from scratch would be quite time consuming
Wow, I was just chiming in with my +1 for the record. Now I look forward to this product even more because talking with the developer at this level makes it even more amazing. Thank you for your responsiveness!
I agree, to write it from scratch would be painful. I was hoping you were running ISC NTPd under the covers and could simply turn on the ntpsnmpd daemon after installing net-snmp and turning on the master agentx socket. But if you're doing closed-source stuff, I'm not sure what to say.
Yes, but if you look closely at that ntpsnmpd it's just a proof of concept and very incomplete, only implementing a few of the MIB objects. hasn't been touched in 8+ years.
Oh, sigh, that's sad. You're right, it is basically an abandoned Google Summer of Code project.
Now that I think of it, my Observium metrics graphing uses a fork of the check_mk agent instead of getting NTP stats over SNMP.
Not that I don't like the NTP250's graphs. I just like to centralize metrics graphing.
But, to get these kinds of stats from an appliance, it might be better to use a standard like SNMP instead of having to support any number of kinds of agents. If I am allowed to keep asking, the next thing I'd ask for is a page that lists stats that Prometheus can scrape (and then that Grafana can easily graph). But I can probably write a transform for the already provided JSONs for Prometheus. SNMP is a lot lower level.
SNMP is now available in v2.7 minus the NTP mib
Wow, thank you! It looks like my device already got the update!
I enabled SNMP, and the device is answering! I already added it to Observium. It is nice to at least have the system resource graphs for now, thank you.
Received the update recently and enabled the feature; thank you for implementing this!
Great new feature, thanks.
How about location/contact and such basic infos?
FYI: Prometheus metrics exporter coming in #101
This is what Observium's Overview page looks like. CPU consumption and CPU load averages might be nice to include just to summarize how much work the machine is doing versus the available hardware it has. I am willing to bet it is mostly bored with an extremely low load.
Would like to see monitoring via SNMP (v2/v3 preferred) as well as minimal ability to configure what it presents (optionally support changing community settings)