My CMTS Monitoring with Grafana and Influxdb | docsis.org

You are here

My CMTS Monitoring with Grafana and Influxdb

5 posts / 0 new
Last post
bajojoba
My CMTS Monitoring with Grafana and Influxdb

Hi guys,

As a docsis engineer with 20 years of experience I would like to present to you the easiest way to monitor your fiber nodes on the CMTS(es). It looks cool and it is big data, so unlimited choises of queries in influxdb exist for you to get the best reports regarding your cable plant. There is some work still to be done for it to be perfect, so I need your help.

https://github.com/l-n-monitoring/CMTS-Monitoring

Please comment and please share!

cmcaldas
monitoring

Hi bajojoba
do you have plans to add 3.0 as well as 3.1 in the future? how much resources does it use on the cmts? I'm thinking the polling and constant gets would cause the cpu utilization to go up. do you know about how much? example 3% or 10%?

thank you for your work and sharing with others

bajojoba
Hi,

Hi,

This is Docsis 3.0. The KPI's are shown per fibernode. Unfortunately I haven't worked on monitoring Docsis 3.1, but I believe that with cablelabs specification it should be easy to implement. I will post more even more information regarding KPI's and a little bit of vendor specific information.

SNMP is a preferred protocol to gather information from CMTSes so there shouldn't be much impact on CPU load. On most of my CMTSes I also monitor CPU load and there are no spikes or higher utilization seen.

If you like it, please give it a star in the github so that it will be higher in search results.

Thanks and BR,
Janko Bajc

JMH
Just wanted to say, this

Just wanted to say, this looks fantastic--would have loved to have had it a couple years ago.

Do you have a list of CMTS' that it works for, or is it fully crossfunctional?
(Motorola BSR, Cisco 10K, Casa 10G, Arris C4, E6, etc)

bajojoba
Thanks

It should be cross-functional. However number of CMs per us/ds is vendor specific. Is a workaround but it requires a lot of resources. There is even much more to this. I've created a dashboard displaying historical/current fibernodes with bad us. You can have a piechart with percentage of good/bad fibernodes etc... Sky is the limit :)