Enabling IPv6 on WRT1900AC causing reboots?

[RogerSC]

I'm beginning to think that my reboot problems with the WRT1900AC are related to having IPv6 enabled, and getting flooded with the Comcast "Neighbour table overflow" messages. I used to get those with the Asus RT-N66U, and was able to login into the router and use a couple of different work-arounds.

Using the WRT1900AC status page, I did see this message in the router log (several times) about the time that the router last rebooted:

May 27 17:02:27 blk-mamba user.warn kernel: IGMP snoop: Local IGMP messages (224.0.0.x) must be flooded

I also see "dhcp6s" running after disabling IPv6 and rebooting, but I'm trying to ignore that *smile*. The "top" output shows 0% of the CPU, but then everything is listed at 0% *smile*.

Other research shows that IGMP snooping is supposed to be a workaround for this, as well. So if this is the problem that I'm seeing, then IGMP and/or the network stack is not handling these messages properly. Maybe the buffer is actually being overrun, and causing the router to reboot, that sort of thing can happen.

I may give Linksys a call and suggest that they look at this. The router should not reboot as a result of this problem, but I could see it being a cause of reboots if not anticipated.

 

[htismaqe]

Those IGMP messages are IPv4 multicast messages. I don't have IPv6 enabled and my provider doesn't even offer IPv6 and I still see those messages once in a while.

On the subject of IGMP and this router, here's something interesting

The truncated log in Sysinfo.cgi shows just ONE normal log message - a wireless client being removed from the Network Map after going to sleep. (Cool, cool feature by the way).

May 27 17:39:34 blk-mamba daemon.notice devidentd[3657]: IP neighbor x.x.x.x [00:00:00:00:0:0] removed (detected via polling)

The other 199 out of 200 log entries look like this:

May 27 17:39:29 blk-mamba user.warn igmpproxy[10259]: No interfaces found for source y.y.y.y

I wonder what enabling "Filter Multicast" in the Security | Firewall page would do?

 

[htismaqe]

OK, I take that back. Those messages are the output of tail -200 /var/log/messages:

If I look at the output of dmesg | tail -200:, I see dozens of messages just like yours.

IGMP snoop: Local IGMP messages (224.0.0.x) must be flooded

Now, in that same output, interspersed with the above, I see these messages:

IGMP Report:239.255.255.250
IGMP Report:224.0.2.3

Going out to IANA and looking up those 2 multicast addresses, I get this:

224.0.2.3	EPSON-disc-set	[SEIKO_EPSON_Corp]	2005-01-01	
239.0.0.0-239.255.255.255	Organization-Local Scope	[David_Meyer][RFC2365]	1997-01-01

Further research on the 239.x net block in RFC 5771 shows this:

10.  Administratively Scoped Block (239/8)

   Addresses in the Administratively Scoped Block are for local use
   within a domain and are described in [RFC2365].

Given all of this, I am going to surmise that these multicast messages are coming from the Epson printer that's connected to my own internal network.

 

[RogerSC]

What I'm thinking is that if I can verify IPv6 as being a factor in the reboots, then I should be looking at the log with IPv6 enabled to see what's there. Now I'm basically speculating based on past history with Comcast IPv6 here, and Neighbour Table overflow problems.

I've got about 18 hours of uptime since the last reboot, now that with IPv6 not enabled. If the router goes past 48 hours without a reboot (a new record *smile*), then I'm re-enabling IPv6, and keeping an eye on the log.

Edit: Now that Victek TomatoARM is available for the R7000, thinking about putting this experiment off for the moment, and trying that out. Should be able to flash directly to that from dd-wrt *smile*.

 

[htismaqe]

The Epson uses Bonjour, which in turn uses mDNS, so that's probably where my multicast messages are coming from.

EDIT: Yeah, there's a MAC address in the log messages, I didn't see it the first time because of the format. It's my printer generating the multicast traffic.

 

[RogerSC]

Hey, Netgear finally fixed my IPv6 problem on the R7000! I can actually reboot or power-cycle the router with IPv6 enabled and still have an internet connection afterwards. Wow, been waiting 8 months for this one.

Hopefully they can keep this working until their next official release. I'm stunned, first time they've actually fixed anything that affects me. I'll be using this debug firmware for a while to make sure that the IPv6 is solid, but this is a great start. Finally!

Update: Still seeing the other IPv6 problem, that my Android phone and iPad 3 don't get IPv6. So, still more to do, but they've at least got a good start on it *smile*.

 

[speedingcheetah]

Hey, Netgear finally fixed my IPv6 problem on the R7000! I can actually reboot or power-cycle the router with IPv6 enabled and still have an internet connection afterwards. Wow, been waiting 8 months for this one.

Hopefully they can keep this working until their next official release. I'm stunned, first time they've actually fixed anything that affects me. I'll be using this debug firmware for a while to make sure that the IPv6 is solid, but this is a great start. Finally!

Update: Still seeing the other IPv6 problem, that my Android phone and iPad 3 don't get IPv6. So, still more to do, but they've at least got a good start on it *smile*.

Is there a particular reason you have ipv6 enabled....?

There really is no advantage to having it enabled at this point. Depends on your ISP, it may actually make things slower if they use a tunnel. I tested having ipv6 on, and to sites that default to ipv6 when it can, tests showed much higher latency. (also, USPS tracking pages did not work with ipv6 on) Things seem faster without ipv6 then with it.

 

[RogerSC]

Is there a particular reason you have ipv6 enabled....?

There really is no advantage to having it enabled at this point. Depends on your ISP, it may actually make things slower if they use a tunnel. I tested having ipv6 on, and to sites that default to ipv6 when it can, tests showed much higher latency. (also, USPS tracking pages did not work with ipv6 on) Things seem faster without ipv6 then with it.

Whether I'm using it or not, it's a router feature that supposed to work. I don't notice greater latency or speed differences with IPv6 when I'm doing stuff on the internet. The ping speeds that I see at my ISP are pretty much the same (13 or 14 ms), no difference that matters to me. Download and upload are also within about 1% of each other. I'm not a gamer, so I'm not trying to wring out the last ms of network latency. I really hate to have broken features on a piece of gear that I've paid $200 for to get extra features (wireless-AC was the important one). In reality, it's mostly on principle that I want things that I'll need in the future to work right.

I've been using dd-wrt for several months, no IPv6 there, but have had hopes of having it working when I need it there as well. If that's the firmware that I'm using when it matters...

http://stage.results.speedtest.comcast.net/result/521183524.png

 

[speedingcheetah]

Whether I'm using it or not, it's a router feature that supposed to work. I don't notice greater latency or speed differences with IPv6 when I'm doing stuff on the internet. The ping speeds that I see at my ISP are pretty much the same (13 or 14 ms), no difference that matters to me. Download and upload are also within about 1% of each other. I'm not a gamer, so I'm not trying to wring out the last ms of network latency. I really hate to have broken features on a piece of gear that I've paid $200 for to get extra features (wireless-AC was the important one). In reality, it's mostly on principle that I want things that I'll need in the future to work right.

I've been using dd-wrt for several months, no IPv6 there, but have had hopes of having it working when I need it there as well. If that's the firmware that I'm using when it matters...

http://stage.results.speedtest.comcast.net/result/521183524.png

For the record....ipv6 on my AC56U (and when I was testing the WRT 1900AC) "worked" since day one. CenturyLink is awesome as far as ipv6 goes...they use 6rd setting. I did extensive testing with ipv4/6 a while back ago and ipv4 was better. I recently turned ipv6 back on and I have not noticed any issues.....maybe they have improved things.

If you have Comcast....good luck. I was told that if u want ipv6 support, you HAVE to have one of their specific modem gateways. Things may have changed and do vary depending on your ISP and area....but don't go blaming the router, when it simple may be your ISP at fault. (granted, I don't know much about your particular situation and what you have tried etc...I am just saying in general here, some routers just don't play well with some isps...I have seen it numerous times)

I have yet to find any info on how to setup ipv6 on Comcast in my area. I did find something, but that was for a tst they dud a few years ago.

Anyway, never trust comcast speed test site...often inaccurate and have been proven to be enhancing the speeds to their own test sites. I ran the comcast speedtest...and it always gives slower results for ipv6..LOL...they say "distance 6400miles"...HA!...I live in the twin cities, about 10 miles from their supposed 'Twin Cities" server. http://stage.results.speedtest.comcast.net/result/521237084.png

Disable ipv6 entirely...and: http://stage.results.speedtest.comcast.net/result/521245484.png

Note: Results are over my AC wifi...a tad faster when I run via wire.

Use independent sites like speedtest.net...etc...there are some ipv6 specific test and diagnostic test sites.

run ICSI netalyzer
that usually can give u an idea of network issues.

also

http://test-ipv6.com/

And you don't have to be a gamer for ipv6 to affect you. In-fact, most game servers are still ipv4. Sites like youtube and other media streaming sites will default to ipv6 if your browser prefers it. If your ISP's implementation of ipv6 is a poorer performer than their ipv4, then you may have issues. I had issues with netfix for a while when ipv6 was on.

 

[RogerSC]

Whether I'm using it or not, it's a router feature that supposed to work. I don't notice greater latency or speed differences with IPv6 when I'm doing stuff on the internet. The ping speeds that I see at my ISP are pretty much the same (13 or 14 ms), no difference that matters to me. Download and upload are also within about 1% of each other. I'm not a gamer, so I'm not trying to wring out the last ms of network latency. I really hate to have broken features on a piece of gear that I've paid $200 for to get extra features (wireless-AC was the important one). In reality, it's mostly on principle that I want things that I'll need in the future to work right.

I've been using dd-wrt for several months, no IPv6 there, but have had hopes of having it working when I need it there as well. If that's the firmware that I'm using when it matters...

To show that there's little interaction between IPv4 speeds here with IPv6 on or off, and how close the Comcast speedtest and speedtest.net results are, here's some more speed test results. The only real difference is the difference in ping time between the Comcast (my ISP *smile*) and the speedtest.net speed tests, not really surprising to me:

Here's my Comcast speedtest iwth IPv6 on:
http://stage.results.speedtest.comcast.net/result/521183524.png

And here's my Comcast speedtest with IPv6 off:
http://stage.results.speedtest.comcast.net/result/521319644.png

Here's my IPv4 speedtest.net result with IPv6 turned off:
http://www.speedtest.net/my-result/3527636331

And here's my IPv4 speedtest.net result with IPv6 turned on:
http://www.speedtest.net/my-result/3527640605

Note that I did reboot the router after turning off IPv6, to make things fair.

 

[sm00thpapa]

Did anyone switch to the WRT1900AC?

Is IPv6 supposed to have an increase in speed? Never heard this mentioned before. If so how does it make it faster? All it is, is a longer IP address?


Sent from my iPad using Tapatalk

 

[RogerSC]

Is IPv6 supposed to have an increase in speed? Never heard this mentioned before. If so how does it make it faster? All it is, is a longer IP address?

Sent from my iPad using Tapatalk

No, IPv6 isn't supposed to increase your speed, it will be a little slower generally. If you look at the Comcast speed tests that I posted, you'll see that the IPv6 download speed is a little slower, about 1%. If nothing else, the IPv6 packet headers are longer, so you're sending more total bits to send the same amount of actual message content bits. Speedingcheetah was saying that under some conditions, if you disable IPv6 that you'll speed up your IPv4. Not the case here, though. Comcast is apparently doing a good enough job that this doesn't happen around here.

 

[sm00thpapa]

No, IPv6 isn't supposed to increase your speed, it will be a little slower generally. If you look at the Comcast speed tests that I posted, you'll see that the IPv6 download speed is a little slower, about 1%. If nothing else, the IPv6 packet headers are longer, so you're sending more total bits to send the same amount of actual message content bits. Speedingcheetah was saying that under some conditions, if you disable IPv6 that you'll speed up your IPv4. Not the case here, though. Comcast is apparently doing a good enough job that this doesn't happen around here.

Ok thank you for the info. I'll have to check my devices to see if they are IPv6 compatible.


Sent from my iPad using Tapatalk

 

[speedingcheetah]

To show that there's little interaction between IPv4 speeds here with IPv6 on or off, and how close the Comcast speedtest and speedtest.net results are, here's some more speed test results. The only real difference is the difference in ping time between the Comcast (my ISP *smile*) and the speedtest.net speed tests, not really surprising to me:

Here's my Comcast speedtest iwth IPv6 on:
http://stage.results.speedtest.comcast.net/result/521183524.png

And here's my Comcast speedtest with IPv6 off:
http://stage.results.speedtest.comcast.net/result/521319644.png

Here's my IPv4 speedtest.net result with IPv6 turned off:
http://www.speedtest.net/my-result/3527636331

And here's my IPv4 speedtest.net result with IPv6 turned on:
http://www.speedtest.net/my-result/3527640605

Note that I did reboot the router after turning off IPv6, to make things fair.

LOL! That's Comcast traffic shaping/prioritization for yea.

I get steady results like that at my folks house when I run multiple tests on thier Comcast.

But come, the evening hours, or late night, when neighbors are home, WAN is at a crawl. Downloads are less than 1MB's and YouTube just buffers...but..run a speedtest from Comcast's or Speedtest.net, and it still says they are getting their full 80/15 service.

"The speedtest is a lie!" direct quote from the last Comcast technician that they called to come and try to "address the issue".

Its funny actually, they have complained so much, and Comcast has come out so many times, and literately replaced everything with new hardware and wires(even the entire box out in their yard, and the one up the street) and upgraded them to their package "Blast" or whatever for free....when it works, Blazing fast...when not, 3g on their phones is faster.

 

[chadster766]

LOL! That's Comcast traffic shaping/prioritization for yea.

I get steady results like that at my folks house when I run multiple tests on thier Comcast.

But come, the evening hours, or late night, when neighbors are home, WAN is at a crawl. Downloads are less than 1MB's and YouTube just buffers...but..run a speedtest from Comcast's or Speedtest.net, and it still says they are getting their full 80/15 service.

"The speedtest is a lie!" direct quote from the last Comcast technician that they called to come and try to "address the issue".

Speedtest.net results will not be real world if the ISP runs their own Speedtest.net server on their network. Then the results are just the transfer rate of ISP internally.

 

[speedingcheetah]

Speedtest.net results will not be real world if the ISP runs their own Speedtest.net server on their network. Then the results are just the transfer rate of ISP internally.

Yup...also, you can run a trace on what server it is connecting to, and see where it is, and then lookup what ISP runs it etc. Did that a few times back when I was messing around with Wireshark.

 

[RogerSC]

LOL! That's Comcast traffic shaping/prioritization for yea.

I get steady results like that at my folks house when I run multiple tests on thier Comcast.

But come, the evening hours, or late night, when neighbors are home, WAN is at a crawl. Downloads are less than 1MB's and YouTube just buffers...but..run a speedtest from Comcast's or Speedtest.net, and it still says they are getting their full 80/15 service.

Same thing here...my internet started slowing down maybe three years ago, until it was 1-2Mbps during the evening instead of the 16Mbps that I was paying for at the time. I kept on Comcast like a cheap suit, and they kept coming out. The techs that came out would tell me that everything was good at my house, there was just a shortage of infrastructure on my side of town. I never got the "customer service" people to admit that, but finally, after about a year, they put in enough infrastructure improvements that I can now get the 25Mbps (28.8, actually *smile*) all day and night. They eventually gave me 6 months of free internet to "compensate me for poor performance", big whoop.

I didn't have the same experience with Comcast and speedtest.net speedtests, though. When the performance felt like 1Mbps, that's what both the Comcast and speedtest.net showed here. Don't know how that would work with speedtest.net, that they would show the full rated speed, unless the speed was normal during the period of time speedtest.net was using to test, and then it dropped after that. I used to hear a lot about Comcast doing that, but I've not experienced it here. When I do a download, there is very little variation in speed, only when the server I'm downloading from gets bogged down. A half-hour download will have a constant rate of accumulation of GB.

The problem now is that in this area we are paying for the Blast! tier, which is 50Mbps everywhere else, 25Mbps here. But again, Comcast has no competition here, so no reason to improve their infrastructure. So our "Performance" tier is 16Mbps instead of 25Mbps, and cost the same as everyone else is paying is paying for Performance, and same with Blast!, half the speed at the same price others are paying for Blast!.

This doesn't engender any love of Comcast from me, as you can imagine...if they had competition here, I'd love to tell Comcast to shove it. Prime example of an unregulated monopoly. Verizon Wireless also acts the same way here, but that's a whole other posting *smile*.

 

[RogerSC]

Yup...also, you can run a trace on what server it is connecting to, and see where it is, and then lookup what ISP runs it etc. Did that a few times back when I was messing around with Wireshark.

Yes, I've checked as well.