Hi,

 Thanks Ilario for sharing, it seems very interesting and make me feel
curios to make some tests with Prometheus. 

 I'm using another stack (collectd, influxdb, grafana) running in a
raspberry pi with docker composer and using the LiMe name resolution
feature to make the nodes send the selected infos to the raspberry from
the image/first boot.

 Since I'm using Luci in my LiMe nodes, I also have luci-statistics for
graphics with collectd and rrd in the node as an double check system.

 At the compilation time I select the extra collectd-plugins that I
need (collectd-uptime, collectd-processes, collectd-ping...) 

 I also put the configuration file from luci-statistics/collectd with
all options  already set (interfaces to monitor, ips to ping, processes
to monitor) and the network output option to send the information to
the machine called monitor.

 So after boot the router is ready and sending packets to the monitor
machine...

 At the raspberry side,  just clone the git, start the docker, change
the hostname to "monitor" and done, all machines are popping up in the
grafana dashboard at monitor:3000 after some minutes

The composer file and folder structure (configs and data) for the
raspberry is here:

https://gitlab.com/rtroian/meshmon/

The readme.md  and the steps are in portuguese, but I believe that the
docker-compose.yml are almost self explanatory.

Abraços e até logo,
Troian


Em qua, 2021-02-17 às 11:33 +0100, amuza escreveu:
> Wow, what a good answer!
>
> Thanks a lot!!
>
> Ilario Gelmetti:
> > Try installing a Prometheus exporter on the nodes and a Prometheus
> > collector (I'm not sure this is the name) on your computer. Then
> > you'll
> > need on your computer also something to plot that data, like
> > Grafana.
> >
> > You can also try with Aparcar's Meshrc, here you can find it and
> > also
> > his bachelor thesis in which it is described:
> > https://github.com/aparcar/meshrc
> > and some discussion
> > https://github.com/libremesh/lime-packages/issues/296
> >
> > Some network profiles used the Prometheus, for example the
> > Valsamoggia one.
> > You can see two files for starting Prometheus exporter at boot
> > here:
> > https://github.com/libremesh/network-profiles/blob/90ae8305d0634f79e2828e8020b350db82f38d2d/es.hackmeeting.org/hm2019generic/root/etc/uci-defaults/99-prometheus_enable
> > and for configuration
> > https://github.com/libremesh/network-profiles/blob/90ae8305d0634f79e2828e8020b350db82f38d2d/es.hackmeeting.org/hm2019generic/root/etc/config/prometheus-node-exporter-lua
> > and the list of the maybe-useful packages (I'm not sure they all
> > are)
> > https://github.com/libremesh/network-profiles/blob/ff2ad943f6a7145c8789a223107845a6a013d67e/es.hackmeeting.org/hm2019generic/Makefile#L4-L7
> >
> > Ciao!
> > Ilario
> >
> > On 2/16/21 6:26 PM, amuza wrote:
> > > Hi,
> > >
> > > Any suggestion to centralize the monitoring (CPU, memory, etc) of
> > > several nodes?
> > >
> > > Preferably with alerts.
> > >
> > > Thank you!
> >
> >
>
> _______________________________________________
> lime-users mailing list
> lime-users@lists.libremesh.org
> https://lists.libremesh.org/mailman/listinfo/lime-users


_______________________________________________
lime-users mailing list
lime-users@lists.libremesh.org
https://lists.libremesh.org/mailman/listinfo/lime-users