'Fatal Signal Six' Rises From the Dead

[Elmer]

GTAXE16000: 3004.388.7_beta1

I recently bought a new router and have been extremely pleased with it, but now after almost flawless operation over the last month, I've had two nights in a row that I woke up to a 'no internet' and this message in the log:

potentially unexpected fatal signal 6.
CPU: 2 PID: 25936 Comm: dnsmasq Tainted: P O 4.19.183 #1
Hardware name: GTAXE16000_2GB (DT)
pstate: 00070010 (nzcv q A32 LE aif)

and a whole bunch of kernel diagnostics (pc, lr, sp, x12 ...).

So, I did a search and saw that this was a problem with AC units in late 2023, but was apparently solved. One possible source of the problem was AiProtection, which I have disabled. Any help would be appreciated.

Thanks

 

[alan6854321]

I started a thread about the dnsmasq crashing issue, did a lot of testing but didn't really come to any conclusion other than there's a problem in dnsmasq itself.

Dnsmasq crashes, watchdog fails to restart it

Hi there, I have a issue where dnsmasq crashes when my WAN connection goes down (If think this is somehow connected to YazDHCP), but it's not really a problem as the watchdog restarts it after a few seconds. However, I've had a few instances recently when the watchdog has failed to restart the...

www.snbforums.com

If you have the scMerlin addon installed, it has an option to watch for the "dnsmasq Tainted:" string and restart dnsmasq when it sees it.

 

[Elmer]

Thanks alan. I thought maybe it was related to some amtm scripts or my storage device, so I deleted all scripts and changed out the storage, then reinstalled the scripts (I am using YazDHCP since I have close to 90 devices). I also added scMerlin and turned on the dnsmasq watchdog, so we'll see what happens. On a positive note, I did not have a crash last night. Thanks again.

 

[alan6854321]

Was your dnsmasq crash associated with anything other events?

E.G. mine always happened after a dnsmasq restart on WAN-down.
But didn't happen after a dnsmasq restart on WAN-up

 

[Elmer]

Not seeing anything related to WAN-down; fortunately the wan connection is pretty stable. Logs so far today are pretty clean. I'll keep watching events and will update with any info. Thx.

 

[Elmer]

Just a follow up on the problem in case someone else stumbles upon it; It has disappeared, so it was probably solved by doing one or more of three things:
1. Installed scMerlin and activated dnsmasq watchdog
2. Removed YazDHCP (YazDHCP and possible storage failure might be intertwined)
3. Possible intermittent storage failure (wasn't sure, replaced the USB drive to be safe)
Sorry I can't narrow it down beyond that broad stroke. Thanks for the help.

 

[langesgnu]

Found this old thread as I was experiencing this issue on my AX88U-Pro with 3004.388.8_2 (or now 8_3). For me the issue was also inconvenient, as directly after a reboot dnsmasq crashed and then also uiscribe, unbound webui and dnvnstat would be gone. If I reinstalled these addons afterwards and did not restart the router, everything was fine and there were no more dnsmasq crahses, but then every reboot would require another installation of the addons.

Now I did a factory reset, then just configured my basic settings, installed YazDHCP and when I either imported my old clients or simply added 1 client manually, then after a reboot immediately dnsmasq crashed with a fatal signal 6. If I remove all clients in the DHCP assignment list and reboot, dnsmasq does not crash.

As per the log, the crash happens after the hostnames file of YazDHCP is read.

May 5 07:05:23 dnsmasq[2347]: read /etc/hosts - 22 names
May 5 07:05:23 dnsmasq[2347]: read /jffs/addons/YazDHCP.d/.hostnames - 1 names
May 5 07:05:23 kernel: potentially unexpected fatal signal 6.
May 5 07:05:23 kernel: CPU: 3 PID: 2347 Comm: dnsmasq Tainted: P 4.19.183 #1

After I uninstalled YazDHCP, rebooted the router, added manual DHCP assignments and rebooted again, there are no more crashes of dnsmasq. So the issue seems to be related to YazDHCP somehow.

 

[Ripshod]

@langesgnu you'll get used to it. If you want to stop it permanently don't use YazDHCP.

 

[bennor]

After I uninstalled YazDHCP, rebooted the router, added manual DHCP assignments and rebooted again, there are no more crashes of dnsmasq. So the issue seems to be related to YazDHCP somehow.

As a datapoint. Not seeing a "fatal error" message using YazDHCP 1.0.7 on a RT-AX86U Pro running 3004.388.8_2. Only scripts running are YazFi and YazDHCP.

To update to YazDHCP 1.0.7 running the following two commands while connected to the router using SSH.

/jffs/scripts/YazDHCP develop
/jffs/scripts/YazDHCP forceupdate

 

[langesgnu]

Thanks for the hint. I gave that a try now but immediately had the dnsmasq crash again, after uninstalling YazDHCP and manually adding the DHCP assignments it is fine for me. Guess I will just leave it without YazDHCP, but maybe others have the same issue.

 

[Ripshod]

I have the same dnsmasq crashes and to be honest it causes no real problems. I do run scMerlin which includes a watchdog that ensures dnsmasq restarts correctly.

 

[JGrana]

Ahh, the very old dnsmasq/YazDHCP tainted issue. I experienced this long ago.
From a lot of debugging, there seemed to be some kind of race/issue with dnsmasq version 2.89 and above and YazDHCP (any version).
For a while, I had a script that would replace dnsmasq 2.89 with 2.85.
Tainted went away.
I now just roll with the latest dnsmasq (2.90 right now) and live with the messages.
Doesn’t seem to have any impact on assigning addresses using YazDHCP’s list.

 

[alan6854321]

I've spent a good while (with help from many forum members) going through this...

Dnsmasq crashes, watchdog fails to restart it

Hi there, I have a issue where dnsmasq crashes when my WAN connection goes down (If think this is somehow connected to YazDHCP), but it's not really a problem as the watchdog restarts it after a few seconds. However, I've had a few instances recently when the watchdog has failed to restart the...

www.snbforums.com

TL: DR - YazDHCP seems to provoke the fault, but there really is a problem with dnsmasq (It throws a segmentation fault) and I don't know how to raise this with anyone who would take me seriously).

The scMerlin "DNS" option provides a work-around most of the time, but fails eventually due to the "nvram lockup" problem.

Stuck commands

Has anyone else observed any "stuck" commands on their router since upgrading to 386.x? Below is a frequent occurrence of the wl command getting stuck being spawned by conn_diag (this process was created at 4:10 AM). But I've also seen it happen with some nvram get commands (I once had Skynet...

www.snbforums.com

So, I wrote my own (Very amateur) script that keeps things going.
BUT, I have a nagging feeling that I really shouldn't have to do any of this.