Is this a bug on Asus firmware? WAN disconnections

[Wutikorn]

I have flashed RT-AC68U with 380.64 over 380.62 on the 20th of December and have not experienced any single WAN drop since then, which is now 5 days. Before there were 2-3 drops every day. I can't see any major fixes or changes to WAN implementation, unless it has been patched by Asus and not mentioned by Merlin in the changelog?

I see neither WAN related issues that get fixed, nor did I had any problem from 380.62. GPL from Asus was not merged between 380.62 and 380.64, only at 380.61. Could it be that there was some error while you were flashing 380.62 at the time? So by flashing any firmware would fix the problem. Or if you are using adaptive QoS, there was or is a problem with both AiProtection and adaptive QoS both on, if I remember correctly.

 

[primitivo]

I see neither WAN related issues that get fixed, nor did I had any problem from 380.62. GPL from Asus was not merged between 380.62 and 380.64, only at 380.61. Could it be that there was some error while you were flashing 380.62 at the time? So by flashing any firmware would fix the problem. Or if you are using adaptive QoS, there was or is a problem with both AiProtection and adaptive QoS both on, if I remember correctly.

For sure not, I have had this issue for quite a while. It is 6th day and still no WAN drop (with 2nd WAN failover based on ping), unless Merlin has removed WAN disconnection prompts from logs to stop people bothering him ;-)

 

[beerkeen]

As an update I am 72 hours in with no WAN drop. So I think it is safe to say that Traffic Analysis is the cause for the WAN drops, at least in my case.

Tell me, please, exactly how you turn off Traffic Analysis?

 

[primitivo]

I had a single WAN drop in 9 last days (since flashing 380.64 over 380.62), which is great. Again, too early to say if something was fixed or changed that was causing this WAN dropping before, but will continue to monitor this.

 

[Scott Jacobs]

I had a single WAN drop in 9 last days (since flashing 380.64 over 380.62), which is great. Again, too early to say if something was fixed or changed that was causing this WAN dropping before, but will continue to monitor this.

Are you using dual wan capability? Or just single? I ask this because I too have updated to 380.64 recently and with single Wan connection I am not seeing these dhcp errors anymore either - Comcast is the provider I am using for my primary connection.

The only time I saw the dhcp connection issues was when I enabled Dual wan in either failover or load balance mode. My secondary provider is Uverse. I have not tried enabling dual wan since switching to 380.64 but may do that in the next day or so to see if the problem is still happening for me.

One thing that is really different with Comcast and Uverse is the expiration interval of the address For Comcast it is 3 days - for Uverse it is much shorter. Don't remember the interval but it seems like the router was having to renew the lease every few hours. Wonder if this disparity is playing a role in the problem that I am seeing with dual wan.

 

[primitivo]

Are you using dual wan capability? Or just single? I ask this because I too have updated to 380.64 recently and with single Wan connection I am not seeing these dhcp errors anymore either - Comcast is the provider I am using for my primary connection.
[...]

I am using in dual-WAN failover mode. My WAN dhcp lease expires every 5 mins... (4G via Dovado bridged), so I have also been suspecting the short renewal time for WAN DHCP drops.

 

[beerkeen]

Tell me, please, exactly how you turn off Traffic Analysis?

Anybody can help me?

 

[Wutikorn]

Anybody can help me?

Not sure if this is what they mean, but you can try going to Traffic Analyzer -> Statistics -> Off.

 

[beerkeen]

Not sure if this is what they mean, but you can try going to Traffic Analyzer -> Statistics -> Off.

I have no this page. Can you show me screenshot? Please.

 

[Wutikorn]

I have no this page. Can you show me screenshot? Please.

It's only for those router with AiProtection and Adaptive QoS. If you don't have either of these, then you probably won't have Traffic Analyzer/Statistic. This might show you an example of the page, but the way to access it is not exactly the same compared to AsusWRT-Merlin firmware. If you still want a screenshot, let me know. I will get send you one when I get home.

 

[beerkeen]

It's only for those router with AiProtection and Adaptive QoS. If you don't have either of these, then you probably won't have Traffic Analyzer/Statistic.

You right. My router is RT-n66U and have no this feature. Sorry.

 

[Pierre Nakashian]

I don't know if this issue is resolved, since this thread has stopped, but I'm still experiencing some minor issues even with 380.66_4.
my ISP DHCP occurs every 5 minutes, I use Putty to remotely connect to my home network over the WAN, if I don't use SSH tunneling feature's my WAN SSH connections doesn't drop, but when I use the Dynamic port forwarding feature of SSH (-D from ssh command line) my remote ssh connection gets disconnected at exactly the time the WAN DHCP renewal occurs. I've tried everything to solve this problem, the only solution so far is to set the RT-AC68u WAN connection type as static, but there are catches with this setup, you have to script it, in case the RT-AC68u reboots the ISP router does not work with the ASUS router configured as static. My scripts are finally stable, just wondering if anyone else is still experiencing this even with the latest firmware/build.

 

[SpAwNtoHell]

Yes. I have this issue also. For info in case someone has ideea about this or something springs to someones mind.

Router Asus RT-AC68U first revision build, no issues till few weeks back, i have updated the stock firmware from GUI, problem persisted like every few days, updated again with a newer firmware... started acting weirder then before wifi also would stop in sthe same time. Attempted to install latest merlin... error message fail... powercycled the router accepted the merlin firmware, but fail to boot. Tried the emergency firmware recovery utility.... router not in recovery mode; tried again and again to get the router in recovery mode... did not work, tried holding wps down and powercycled, first time it looked it done something, could not repeat the same behaviour router still not in recovery mode... convinced after a week that bootloader is damaged i almost gived up without doing anything just power on entered in recovery mode... only accepted a firmware stock from almost 2 years back... oh well working now i thought... after five minutes traffic stoped through wan port.... atempted to flash with merlin... worked this time... all fine i thought but 4 days later internet dropped ... again the dhcp error message was displayed by merlin firmware... my wife contacted the isp, they not offer support for connection if is not their equipment they sent over next day a D-link 3782... with this as a 2 in 1 eliminated my vdsl modem also...24h later internet dropped ... took it of put my modem on and bought a rt-ac66u b1 as a desparate measure 48h and seems ok but not holding my breath... on isp router or the modem i have i cannot see any errors when connection drops and the leds work as it should. Something that i forgot to mention with th rt ac68u if i unplug the modem to wan patch cable for 1 sec and plug right back internet is up again till next time it happens. The patch cable is cat5e as everything around my place so not related. The distance to exchange is 0.45miles. Never had issues till now for 4+ years same isp ,router ,cable, modem tho i think something changed at their end i am very confused as the d-link showed similar issues in connection drop but cannot be 100% sure as i cannot see wan diagnostics on that device. Waiting for a draytek vigor 130 to see if it makes a difference...

 

[Pierre Nakashian]

I took a little peak inside RT-AC68U during WAN dhcp requests, my WAN SSH connection to the router
got dropped almost every 5 minutes when I received a UDP packet from the ISP provider
(when I have dynamic port forwarding). I monitored DHCP requests using the counts in iptable
with spt:67 dpt:68.

I noticed a few extra arguments to the udhcpc process handling this request.

/sbin/udhcpc -i eth0 -p /var/run/udhcpc0.pid -s /tmp/udhcpc -O33 -O249

/tmp/udhcpc is actually a symbolic link to /sbin/rc

the process udhcpc is handling incoming DHCP request, then runs a linux command
/sbin/rc -O33 -O249

googling source code I found, this code snippet, this is outside my expertise I think routes are possibly being affected partially
during DHCP requests. this event does not disconnect my WAN SSH connection with no tunneling,
just SSH connection with at least Dynamic tunneling sometimes known as SOCKS proxy.
notice the same -O33 and -O249 numbers in the code snippet below


https://github.com/RMerl/asuswrt-merlin/blob/master/release/src/router/rc/udhcpc.c

/* 0: disable, 1: MS routes, 2: RFC routes, 3: Both */
dr_enable = nvram_get_int("dr_enable_x");
if (dr_enable != 0) {
dhcp_argv[index++] = "-O33";/* routes */
if (dr_enable & (1 << 0))
dhcp_argv[index++] = "-O249";/* "msstaticroutes" */
if (dr_enable & (1 << 1))
dhcp_argv[index++] = "-O121";/* "staticroutes" */
}


I'll have to experiment with this new clue,

I manually killed the udhcpc process so the "/sbin/rc" would not run indirectly anymore,
seems my SSH with dynamic tunneling seems stable.

I'll have to also try running udhcpc with different arguments (minus the /sbin/rc ...) and setting up routes myself,
since my routes are the same every day, I can also try the nvram setting as well, "dr_enable_x" is set to 1
for WAN connection type configured as Dynamic.

 

[RMerlin]

/* 0: disable, 1: MS routes, 2: RFC routes, 3: Both */

It might not be obvious (you would have expected this to be on the WAN page), but this setting is configurable on the LAN -> IPTV page of the webui.

 

[Pierre Nakashian]

thanks Merlin,

I finished testing. with the exception of killing UDHCPC process nothing else changed the behavior. So I've come up with a simple compromise, I've written a script and placed in "/jffs/scripts/service-start" to kill the UDHCPC process during startup after a 10 min delay. the script will continuously check if internet access is up, when it's not it will try re-running the UDHCPC process and exit. When that doesn't solve the problem, the script will reboot the rooter after 2 minutes of no internet access.

The reason I chose this solution, is My IP address never changes even though I don't have static IP service. during reboot my ISP hardware (AT&T) detects if my router is configured as STATIC Wan Connection type and It blocks internet access. This compromise allows my router configured in Automatic Wan Connection type but behave as if I have it configured as STATIC Wan Connection type.

I did leave out this parameter -s /tmp/udhcpc -O33 -O249
from the UDHCPC command in the script, I'll have to test and wait to see if there is any impact from that.
also my router is not configured with Dual Wan connection, the script assumes only 1 WAN interface, it will
use the NVRAM get command whether your WAN interface is either eth0 or vlan2.

here's the script.
cut between lines and save to file named at this location
then run
chmod +x /jffs/scripts/services-start

/jffs/scripts/services-start
-----------services-start--------------------------------------
#!/bin/sh
sleep 600
killall -9 udhcpc

export COUNT=0
export ETH=`nvram get wan0_ifname`

while true
do
`ping -c 4 www.yahoo.com > /dev/null 2>&1`
RESULT="$?"

if [ $RESULT -ne 0 ]
then
COUNT=`expr $COUNT + 1`
if [ $COUNT -ge 12 ]
then
/sbin/reboot
elif [ $COUNT -eq 2 ]
then
#only run udhcpc when internet access seems to be down, around 20sec + ping time
#handle 1 dhcp request from ISP, after obtaining lease gracefully exit
/sbin/udhcpc -q -f -i $ETH -p /var/run/udhcpc0.pid
fi
sleep 10
else
COUNT=0
sleep 60
fi

done
----------------------------------------------------------------

 

[shdwkeeper]

Is this still the same issue that is happening in this post?

http://www.smallnetbuilderforums.com/threads/rt-ac87u-keeps-locking-up.27283/page-4#post-240615

 

[ColinTaylor]

Is this still the same issue that is happening in this post?

http://www.smallnetbuilderforums.com/threads/rt-ac87u-keeps-locking-up.27283/page-4#post-240615

I can't see any connection with this thread and the link you posted. It seems to be a completely different problem.

 

[lev]

thanks Merlin,

I finished testing. with the exception of killing UDHCPC process nothing else changed the behavior. So I've come up with a simple compromise

Thanks much for sharing your solution. Going to try it as I have the same issue.

 

[shdwkeeper]

I can't see any connection with this thread and the link you posted. It seems to be a completely different problem.

Wan disconnection and and having to reboot the router sounds pretty similar to me, but maybe you didn't read entire post in link above. Cheers.