Collectd ping; ICMP drop rate question

I'm experimenting with luci-app-statistics, specifically the ping plugin. I have the plugin enabled to monitor two client 192.168.1.50 and 192.168.1.60. This results in the data below.

What caught my eye was the ICMP drop rate. I have two questions about this:

  1. What exactly is the ICMP drop rate? Is it related to each client or the network as a whole?
  2. What could be the cause that the ICMP drop rate starts to increase at around 16:00 (for both 192.168.1.50 and -.60) and climax at around 5:00 when 192.168.1.60 fails and 192.168.1.50 seems fine again?

That graph shows the percentage of ICMP packets sent by the router that where not answered by the client. This van be caused by a congested network, an unresponsive client, or many other similar causes.

2 Likes

Let assume its congestion; As both client show the same drop rate (up untill 05:00) can I assume that another client on the network is the cause of the congestion or is that a wrong assumption?

Are they wireless clients or cabled?
Maybe the .60 got disconnected at 05:00 and didn't respond to pings at all.
It doesn't need to be another client that causes the congestion. If either of these 3 monitored clients starts downloading or uploading too much the response times are expected to rise.

That graph is about drop rate, not response time (latency). The share of ping packets getting lost (= no answer).

First guess might be some network activity like torrenting, streaming or so, combined to SQM cake drop packets for maintaing quality (if you have configured SQM to have impact on LAN). Somebody came home at 16:30 and started networking?

And at 04:45 a device shut itself down and became unreachable. Large download completed and the client was configured to shut the PC down? Or some smart powersave action from a device like "go to deep sleep after 6 hours of inactivity).

My network isn't that sofisticated. I have a R7800 as my main router on standard Openwrt (19.07.2). No additional package like sqm, etc. Another R7800 as dumb AP is connected via copper.
Both the .50 and .60 are (wifi) cameras and they are streaming 24/7 video to my NAS (connected via copper to router).

The .60 shows frequent disconnects which I can't explain. It has a much better SNR than the .50, but the .50 has a stable connection over many weeks, the .60 disconnects almost daily (like at 5:00).
I'm looking for a reason why this is happening and I thought monitoring the ping would be a good place to start.

You are right in that probably around 16:00 network activity increases but this certainly does not continue past 00:00.

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.