[Fork] Asuswrt-Merlin 374.43 LTS releases (Archive)

[john9527]

dnsmasq dies immediately.

Please define this.....what does 'die' mean?

what is the output of
ps | grep dnsmasq

what is the last log of dnsmasq in the syslog?

execute the following which will dump some data to the syslog....what does it say?
/usr/bin/killall -s USR1 dnsmasq

 

[slobodan]

Cannot reply here, see http://members.ziggo.nl/slobodan/reply.txt

In case one wonders about the bad entries, here they are:

192.168.2.2 ▒supportjy2xvvdmx.onion
::1 ▒supportjy2xvvdmx.onion

The character is 92 (hex) or 146 (decimal).

 

[jrmwvu04]

Cannot reply here, see http://members.ziggo.nl/slobodan/reply.txt

In case one wonders about the bad entries, here they are:

192.168.2.2 ▒supportjy2xvvdmx.onion
::1 ▒supportjy2xvvdmx.onion

The character is 92 (hex) or 146 (decimal).

Well I'll ask the silly question - have you tried removing those lines and restarting dnsmasq? Three quarters of a million hosts seems pretty intense.

 

[ColinTaylor]

@slobodan

Well I'll ask the silly question - have you tried removing those lines and restarting dnsmasq?

Exactly what I was thinking. The other obvious thing would be to disable dnscrypt. That would establish if either of those things is connected with the problem.

 

[john9527]

Cannot reply here, see....

Well, the two command runs you did at the end show dnsmasq is alive and processing DNS requests.....
Can you upload a full syslog from boot start and PM me a link?

 

[slobodan]

Well I'll ask the silly question - have you tried removing those lines and restarting dnsmasq? Three quarters of a million hosts seems pretty intense.

Yes, it is the same without those two lines.

 

[slobodan]

Well, the two command runs you did at the end show dnsmasq is alive and processing DNS requests.....

Because I have entered manually dnsmasq --log-async . However, it was not alive for long, even after that.

Can you upload a full syslog from boot start and PM me a link?

I will do it later.

 

[john9527]

Because I have entered manually dnsmasq --log-async . However, it was not alive for long, even after that.

Argh....makes it hard to do debug when you 'adlib'

 

[ahduck]

Hi John,
Just updated my AC68U to v25E1.
Everything seems fine in Router mode. But whilst trying out in Wireless Bridge mode, l'm not able to save any changes made (keep processing...) cannot clear logs. or reboot router in GUI using IE or Firefox.
Need your advise.. Thanks

 

[slobodan]

I have found the dnsmasq killer: it dies as soon as I ssh to the router (but not only then, alas).

 

[john9527]

Hi John,
Just updated my AC68U to v25E1.
Everything seems fine in Router mode. But whilst trying out in Wireless Bridge mode, l'm not able to save any changes made (keep processing...) cannot clear logs. or reboot router in GUI using IE or Firefox.
Need your advise.. Thanks

Turn on the browser console (F12) and look under the console menu bar selection for any errors (things of interest are usually highlighted in red).

 

[john9527]

I have found the dnsmasq killer: it dies as soon as I ssh to the router (but not only then, alas).

Actually, in the syslog you made available, the problems started long before you logged in. Something is hanging the startup so that a bunch of processes get put in a wait state.....it's either radvd for your 6in4 tunnel (don't think it's this one) or dnscrypt. With respect to dnscrypt...have you

• updated the server listing? run 'update-dnscrypt-resolvers.sh' from the command line (I have this on a cron job that runs once a week)
• Are you sure your dnscrypt servers aren't in your blocking_file? There is the possibility that there is a race in initially bringing up dnscrypt and dnsmasq....and if dnsmasq wins with the server in the blocking file, you are dead from the get-go.....otherwise, things will initially come up, then die quickly.

In any event, try with dnscrypt disabled.

 

[R1-Limited]

@john9527
Sorry did not correct the issue for me. I have syslogs from both if needed. I have reverted back to 24E3

 

[john9527]

@john9527
Sorry did not correct the issue for me. I have syslogs from both if needed. I have reverted back to 24E3

yes....please upload them to a file sharing site and PM me the links.

 

[slobodan]

Actually, in the syslog you made available, the problems started long before you logged in. Something is hanging the startup so that a bunch of processes get put in a wait state.....it's either radvd for your 6in4 tunnel (don't think it's this one) or dnscrypt. With respect to dnscrypt...have you

• updated the server listing? run 'update-dnscrypt-resolvers.sh' from the command line (I have this on a cron job that runs once a week)

Did it, did not help.

• Are you sure your dnscrypt servers aren't in your blocking_file? There is the possibility that there is a race in initially bringing up dnscrypt and dnsmasq....and if dnsmasq wins with the server in the blocking file, you are dead from the get-go.....otherwise, things will initially come up, then die quickly.

In any event, try with dnscrypt disabled.

Actually hosts block domain names, they don't block IPs. dnscrypt uses IPs for connecting to server, not domain names.

But it seems to be tied to dnscrypt, dnsmasq works fine without dnscrypt.

 

[john9527]

But it seems to be tied to dnscrypt, dnsmasq works fine without dnscrypt.

Have you ever installed dnscrypt via entware? or an entropy generator (haveged or rng-tools)?

 

[ahduck]

Turn on the browser console (F12) and look under the console menu bar selection for any errors (things of interest are usually highlighted in red).

This is all I got.

 

[slobodan]

Have you ever installed dnscrypt via entware? or an entropy generator (haveged or rng-tools)?

No, not on this router. I have an RT-N16 which previously had dnscrypt installed from entware, but it works fine with 25E1.

 

[john9527]

NEW TEST BUILD - V25T3
https://1drv.ms/f/s!Ainhp1nBLzMJkEFTnFp4Sh2wvQva

@R1-Limited
Your syslog shows a completely different signature. At least part of what is going on was supposed to be fixed by backing out an RFC in the later builds. I'm now wondering if the RFC was just a way to trigger another problem in the code????

So, I cherry-picked a select set of fixes from 2.77 and made a custom 2.76fork dnsmasq. These fixes should be pretty safe and address the TCP retry problem I was having on Linux plus a couple of other problems that seemed should be addressed.

@slobodan
On a hunch, I made a couple of changes to the startup sequence for dnscrypt/dnsmasq. I'm thinking that loading that huge hosts file may be interfering with the dnscrypt start.

@Lotta Cox
You are always welcome to give this a try as well (same for anyone else).

 

[Lotta Cox]

Installed T3
T2 was perfectly stable for 2 days 1 hours 46 minutes 43 seconds