Gateway Monitoring and Packet Loss

[Meg]

Hello: I recently started monitoring my gateway and noticed that I am getting intervals of packet loss. I am running opnsense 25.7.7_4 with adguard home and unbound as my recursive resolver. I am also using zenarmor. I was just wondering if anybody can explain what I am seeing here on the health/quality graph and what could be causing it. As I have never monitored this before I am not sure if this is normal behavior. See attached graph.

[keeka]

The default monitor address (derived from the gateway) may fail, or be slow, to respond to ICMP requests. Try changing the monitor IP in the gateway settings.
The delay also looks rather large IMO. If you disable the 'loss' series in that graph, you will get a finer grained view of the delay (RTT). I expect that will also improve if you change the monitor IP.
Of course none of this will improve your connection but it may give a better indication of the real state of the link.

[meyergru]

ICMP is a low-priority service that is not guaranteed to work. Some gateways do not reliably answer to those requests, especially when they are under high load.

[Meg]

Thanks for the replies. I changed my monitor ip to google and cloudflare  which both still had somewhat high delay and still some packet loss to google. I switched to quad9 and my delay has dropped from an average of 33ms with occasional  spikes to 66ms and still some packet loss wsith 8.8.8.8 down to 22ms with occasional spikes to 32ms with no packet loss on quad9. So it looks like its monitoring more accurately now.

[SenseX]

Hi,

How do I monitor the RTT of the gateway in Reporting:Health?
My RTT is around 2.5 ms, and I would like to see a graph of this on the reporting page, but I can't figure out where to look.

[OPNenthu]

I think you have to first enable the RRD graphing backend in Reporting->Settings, then after a few minutes look under Reporting->Health->Category="Quality" and select the GW.

@Meg, are you on a cable modem line by chance?  I am and my charts look very similar.  The only "clean" graph I've been able to get is from the OPNsense instance at my parents' house which is on a fiber line.

DOCSIS/cable and CMTS systems are quite variable from what I've seen with my home connection.  Service fluctuates between 500-1200 Mbps throughout the day, making SQM difficult to nail down.  Just this week my ISP was doing upgrades in my area and I think I got moved to a different CMTS because my prefix finally changed and my baseline latency in speed tests went up (from 14 to 20 ms).

The CMTS sometimes doesn't respond to pings and I've had to instead monitor the 2nd or 3rd hop router.  I just gave up and disabled WAN monitoring entirely since I only wanted the graphs.   I would say that unless failover is required, cable modem lines just aren't worth monitoring.  It's too noisy.

Am curious to hear from other cable subscribers about this.

[SenseX]

Hi,
It has been enable all the time. It's look like it some decimal problem here 0.002, should be 2.3ms or so :)