Ultra lightweight config for Raspberry Pi

Hi, I’d like to use Netdata in a Raspberry Pi 2, but its default installation feels heavy, as it collects a LOT of metrics per second.

What setup/config would you recommend for a Raspberry Pi? Consider

  • Being gentle with writes to disk, as RPi uses an SD card that have a limited lifespan and sometimes get corrupted. I know that Netdata supports RAM storage, but it’d nice to dump from RAM to disk from time to time, so that there’s historical data.
  • The minimum and most important metrics (cpu, mem, disk, net in/out)

EDIT: being lazy reading the docs. Now reading this Monitor Pi-hole (and a Raspberry Pi) with Netdata | Learn Netdata

Anyway, tips are welcome

Hello @redraw ,

You can change your update every parameter inside netdata.conf, disable some plugins inside the same file. Or a combination of both.

Best regards!

I read dbengine provides 3 storage tiers, with different update every values. Each one aggregates the previous one, so that’s ok. I was afraid that 1s resolution would turn big, but it’s not a big deal as it’s limited by dbengine multihost disk space MB and aggregated on the other tiers.

However, it’s a bit confusing which plugins to turn on/off, which are important, and which optional. For example, I disabled the apps one, and it seems to have reduced the metrics collected per sec, but now there are more charts, one for each systemd service. I guess it’s a matter of spending some time tinkering…

Why if I set update every 5 inside [db] block and restart netdata, I correctly see “Granularity: 5s” hovering a chart, however I keep seeing 1s sample rate in the dashboard live?

Systemd and services are part of cgroup.plugin.

I configured my netdata.conf as:

[db]
        update every = 5

And I am having the external collectors running with 5 seconds:

# ps aux| grep netdata
netdata  16200  0.0  0.0  16184  6976 ?        SN   13:59   0:00 /usr/libexec/netdata/plugins.d/debugfs.plugin 5
netdata  16205  0.3  0.0 170816 12992 ?        SNl  13:59   0:03 /usr/libexec/netdata/plugins.d/apps.plugin 5 with-function-cmdline
netdata  16212  0.0  0.3 1295280 56660 ?       SNl  13:59   0:00 /usr/libexec/netdata/plugins.d/go.d.plugin 5
netdata  16213  0.0  0.0  16040  6824 ?        SN   13:59   0:00 /usr/libexec/netdata/plugins.d/perf.plugin 5 cycles
netdata  16215  0.0  0.0  16180  6932 ?        SN   13:59   0:00 /usr/libexec/netdata/plugins.d/slabinfo.plugin 5
root     16218  0.0  0.0  16084  6748 ?        SN   13:59   0:00 /usr/libexec/netdata/plugins.d/nfacct.plugin 5
netdata  16238  0.0  0.0   4412  3372 ?        SN   13:59   0:00 bash /usr/libexec/netdata/plugins.d/tc-qos-helper.sh 5
root     16245  0.4  0.5 1505844 83388 ?       SNl  13:59   0:03 /usr/libexec/netdata/plugins.d/ebpf.plugin 5

Which option did you change in our host?

Best regads!

yep, I have exactly the same.

thing is I’m not sure how charts can show a 1s resolution, unique values, is it interpolating or something?