RT-AC86U poor DNS response times

[Adrian Knight]

Getting poor DNS response times from the RT-AC86U max time is 273ms with an average of 70ms.

I have setup a probe to my ISPs DNS servers and they are responding on average of 13ms, so it's defiantly an issue with the router.

 

[ColinTaylor]

Pictures are too small to read.

 

[Adrian Knight]

I have a high screen resolution and unfortunately it's the way this forum sw scales them, but you can use browser zoom to be able to read them.

 

[ColinTaylor]

Even with browser zoom some of the text is illegible and the rest barely legible.

 

[Adrian Knight]

Added as thumbnails, so you should be able to click and download now.

 

[kfp]

Do you use DNS based or hosts file based blocking?

 

[Adrian Knight]

Do you use DNS based or hosts file based blocking?

Nothing like that at all.

 

[ColinTaylor]

If you issue the following command on the router and then look at the syslog you might see a problem.

killall -s USR1 dnsmasq

Jun 27 19:58:57 dnsmasq[29373]: time 90258
Jun 27 19:58:57 dnsmasq[29373]: cache size 1500, 0/6942 cache insertions re-used unexpired cache entries.
Jun 27 19:58:57 dnsmasq[29373]: queries forwarded 1730, queries answered locally 1477
Jun 27 19:58:57 dnsmasq[29373]: server 194.168.4.100#53: queries sent 1377, retried or failed 5
Jun 27 19:58:57 dnsmasq[29373]: server 194.168.8.100#53: queries sent 711, retried or failed 11

 

[Adrian Knight]

If you issue the following command on the router and then look at the syslog you might see a problem.

killall -s USR1 dnsmasq

I do not see anything in syslog after using that command, have also tried SIGUSR1

 

[ColinTaylor]

I do not see anything in syslog after using that command, have also tried SIGUSR1

Sorry, no idea. Must be some difference between stock firmware and Merlin's. Perhaps the log level is set differently.

 

[kfp]

What’s your setup for testing with and without router pings?

There’s a brief moment where latency dropped to your normal numbers before stopping, what happened there?

BTW, screenshots are still hard to read.

 

[Adrian Knight]

What’s your setup for testing with and without router pings?

There’s a brief moment where latency dropped to your normal numbers before stopping, what happened there?

BTW, screenshots are still hard to read.

These links should help.
https://ibb.co/ku1jx8
https://ibb.co/gBMWAT

The drop is when my laptop was turned off whilst I was sleeping

 

[kfp]

OK, that’s way better. So obvious follow up questions would be: What’s on 4.100 and 8.100? Are they both running dnsmasq and forwarding to the same DNS servers? Who’s doing the DNS quested? I’m guess your laptop and to five different servers?

Asked about the gap because values dropped close to the normal range before the data stopped.

 

[ColinTaylor]

@kfp Look closely . It's 194.168.4.100. They're Virgin Media's DNS servers.

Still no idea why it's so slow for him .

 

[kfp]

@kfp Look closely . It's 194.168.4.100. They're Virgin Media's DNS servers.

Good catch, I demand more pixels!

If the router is forwarding to the same DNS servers I would think that it’s probably some odd dnsmasq configuration or something is pinning the CPU. Factory reset might help but I’m out of ideas as well unless there are more information..

 

[sfx2000]

Still no idea why it's so slow for him .

IIRC, PRTG uses SNMP, so it might just be a logging task thing with the lib - SNMP and Syslog are low priority in the overall scheme of things...

 

[kfp]

IIRC, PRTG uses SNMP, so it might just be a logging task thing with the lib - SNMP and Syslog are low priority in the overall scheme of things...

SNMP is used for retrieving system stats, but probably not for measuring DNS latency. Also, I’d be very surprised if OpenDNS and most ISP DNS servers let you pull anything via SNMP.

 

[sfx2000]

SNMP is used for retrieving system stats, but probably not for measuring DNS latency. Also, I’d be very surprised if OpenDNS and most ISP DNS servers let you pull anything via SNMP.

Agreed - but PRTG can pull status from the Router

 

[kfp]

Agreed - but PRTG can pull status from the Router

The arrival of said data might be lower priority but shouldn’t impact the recorded ms numbers.

Though I have to admit QoS never crossed my mind

 

[sfx2000]

The arrival of said data might be lower priority but shouldn’t impact the recorded ms numbers.

Though I have to admit QoS never crossed my mind

Not QoS for snmp traffic over to the collector - more task priority for the snmp agent on the device (in this case, the Asus router) - and depending on the log mask there, can be some latency... not everything runs in real-time on embedded Linux, important tasks do, SNMP and Syslog are not important.

One has to look at trends over time...