384.19 DNS very slow

[ColinTaylor]

Have you tried any other versions of the firmware?

Did you have this problem with previous versions of Merlin?

 

[Nodiaque]

Did you have this problem with previous versions of Merlin?

I don't recall

 

[ColinTaylor]

Given that you've already factory reset your router I'd be inclined to try a different firmware version, either 384.18 or 386.1_beta4.

 

[dave14305]

Post the contents of /tmp/resolv.dnsmasq and try the nslookup with each listed server individually like you did with 8.8.8.8. Maybe one of your ISP DNS servers is unreachable.

 

[shabbs]

I don't recall

What entries are showing for your WAN DNS when you have it set to your ISP? See here:

 

[cptnoblivious]

Post the contents of /tmp/resolv.dnsmasq and try the nslookup with each listed server individually like you did with 8.8.8.8. Maybe one of your ISP DNS servers is unreachable.

Even if one of them is unreachable, the router should just switch to the other server. At least that's the understanding I have about how the client rotates between the servers.

 

[joe scian]

No didn't try to go back to oem firmware. Factory Reset was to be sure merlin was properly configured since I always did in-place upgrade without configuration reset.

Whats your Router CPU doing when you are getting the slow response. Is it CPU Bound ? Whats the top 5 CPU processes in htop? I assume you have DNS Filter set to Router. When you ping your ISP DNS whats the response ? Fast ?

 

[john9527]

Do you have DNSSEC enabled? Most (all?) ISP DNS servers are not DNSSEC enabled or are misconfigured for DNSSEC. Google DNS idoes support it.

 

[joe scian]

Do you have DNSSEC enabled? Most (all?) ISP DNS servers are not DNSSEC enabled or are misconfigured for DNSSEC. Google DNS idoes support it.

NO he does not - from post No. 8

 

[dave14305]

Your router LAN IP is 192.168.0.2, but you have another ASUS device with a reservation for 192.168.0.1? Is there anything else peculiar about your LAN setup?

 

[john9527]

@Nodiaque
ssh to the router and run the following command when you are having the problem

/usr/bin/killall -s USR1 dnsmasq

then post the stats it will have written to the syslog

@joe scian
Missed the screen shot...was a bit too small on the screen I was using.

 

[joe scian]

@Nodiaque
ssh to the router and run the following command when you are having the problem

/usr/bin/killall -s USR1 dnsmasq

then post the stats it will have written to the syslog

@joe scian
Missed the screen shot...was a bit too small on the screen I was using.

I thinkI know what his problem is - He has DHCP Option 42 set as 0.0.0.0. dhcp-option=lan,42,0.0.0.0 . Surely this should be set to his Router IP Address - ie. dhcp-option=lan,42,192.168.0.2
Option 42 (NTP Server)
DescriptionThe NTP server option specifies the NTP server (RFC 868), available to the client. The code for the NTP server option is 42. The minimum length for this option is 4 octets, and the length MUST always be a multiple of 4.

You might also try ipconfig -flushdns from cmd prompt.

Also an observation only - from his DHCP reservations - ALL of his devices are assigned manually outside of the DHCP reservations which are set at 192.168.0.100 to 192.168.0.150. And I assume he has CGNAT WAN Connection instead of PPPOE since he is using Automatic IP ?

BTW John - my dnsmasq cache is set to 0 and dosent change when i enable it to 1500 in Diversion. Also set it manually to 1500 in etc/dnsmasq.conf - restart dnsmasq and setting goes back to 0. I assume this is because I am using Unbound and Unbound cache overides this to zero. If DHCP option 42 is set at 0.0.0.0 then that may be because he has set a specific NTP Server and enabled NTP Server in Admin/system. That could negate my comments above - but worth trying at any rate.

 

[ColinTaylor]

@joe scian There is nothing wrong with that dnsmasq NTP option:

The special address 0.0.0.0 is taken to mean "the address of the machine running dnsmasq".

It's also not relevant to his problem which is DNS resolution and not what DHCP options are pushed to the client.

There's also nothing wrong with his DHCP reservation addresses. There's nothing to suggest he's behind CGNAT and even if he were that wouldn't cause his problem.

 

[Nodiaque]

Sorry for late reply, I have to check why I don't receive email notification.

The slow DNS is always, it's not sometime, it's always. CPU is doing either nothing or working depending on when I check, but it's not CPU problem it seems.

Option 42 is for NTP server, which is for time server. Time server isn't enabled on the router since Windows 10 already get his time from MS and between computer on the LAN.

As for the DHCP reservation, this is how you're suppose to do it. You don't reserve IP in the dynamic range, this can cause dual IP on the lan. Best practice is to have a dedicated range that you do static dhcp reservation in it. All my "servers" and PC are below 100, and over 150 are my iot devices.

Flushdns doesn't have anything to do with it, this simply flush the DNS cache on the client. Since all client have this problem, irrelevant (and since it's windows, you can bet they restart at least once a month).

Why are you saying I have another Asus device on .1? This is a PC with an asus motherboard.

Again, the issue isn't my ISP DNS. If I configure the DHCP to send the same DNS ip to my computers instead of the router DNS, everything goes smooth as butter. But once I revert to my router, we have delays.

My Internet is cable internet (coax) with a modem DCM475.

I'm thinking either my router or the firmware is having a hard time. I have other problem for the past few months where all my connected device (wired or wireless) loose up to 4 ping to google and on the router side, the same ping is >1000ms. This cause my VPN and streaming to crash since it's many seconds. This is totally random and have nothing to do with the current DNS issue. I'm thinking about flashing again to asus stock and see if everything go back to normal. Right now for troubleshooting of this problem (the connection lost), I have connected my main pc directly to my modem and connected the wan port of the router to my second nic where I share my internet. The pc connected directly to the internet have no problem but the 2 other have something ping lost, but not the router (I'm now pinging 1 wired on the router, 1 wifi on the router, the router and the pc connected directly to the internet).

right now, since I'm connected through my PC, I won't print what was asked since it won't reflect current reality since I'm in front of the router, not behind.

 

[dave14305]

Why are you saying I have another Asus device on .1? This is a PC with an asus motherboard.

Because your dnsmasq.conf had this entry:

dhcp-host=E0:CB:4E:F9:6D:FE,set:E0:CB:4E:F9:6D:FE,192.168.0.1

And looking up that MAC suggested ASUS as the manufacturer E0:CB:4E ASUSTek COMPUTER INC. If that's a PC, fine by me.

Since you've gone through some effort to reconfigure your LAN so the router is .2 instead of .1, I was wondering what else you may have done that is not typical for an out-of-the-box ASUS router.

 

[Nodiaque]

I'm wondering what you call an effort... By default, the router is .1.1 and I switch to 0.2, not a big deal. The dhcp is changed automaticaly to that range and I shrink the range. all remaining is as merlin decided in his firmware. It's far from being such a hard job that require effort, redoing each dhcp reservation is something that require some effort. Changing 4 numbers in a field...

 

[L&LD]

And the question @dave14305 asked you is still not answered.

 

[Nodiaque]

What question are you referring too? I'm looking and I don't see any question that isn't answered

 

[dave14305]

I'm wondering what you call an effort...

Trying to help you has become an effort. Wishing you the best of luck with solving your problem.

 

[shabbs]

@Nodiaque

What entries are you seeing in this screen for your DNS? (please blank out your WAN IP and Gateway info like I have, no need to share that)