 | [Speed] PSA: Slow WiFi on Comcast? May be Comcasts Fault There is a critical bug in Comcast's network configuration that is likely crippling thousands of routers and pretty much only Comcast has this issue. It is an edge case interaction caused by bad DSCP information coming from Comcast and poor condition handling by the WMM driver on certain linksys/cisco eseries routers and possibly others as well. This issue can appear on both stock and modded routers. Comcast's network is misconfigured and has been for years and has likely caused many people to needlessly replace fully functional routers, IPv6 speeds always worked for me because DSCP was configured correctly for that but it is incorrect for IPv4. Since the only packets with this bad information are the ones that are being downloaded upload is not affected nearly as badly. This can affect any router that uses QoS or WMM(every wireless N router has WMM)
This is the fix I made and used on an e2000 and e3000 running shibby 112:
put this in init scripts section: insmod xt_DSCP.ko
and this in wan up scripts: iptables -t mangle -A PREROUTING -i vlan2 -j DSCP set-dscp 0
these may need to be modified slightly depending on configuration.
This corrects the bad DSCP header information as soon as a packet hits the WAN interface on a router. DSCP as received from Comcast's network is 0x08 this is the lowest priority possible which WMM interpertes as being unimportant to transmit quickly. The scripts listed changes the DSCP on all packets to 0x00 which essentially means unclassified which WMM handles properly. This is how virtually every other ISP uses DSCP and is why the issues is specific to Comcast. I have confirmed this issue on connections in both Chicago and Boulder, CO and it is likely to be present on Comcast's entire network. Cisco appears to have at some point released new WMM drivers for stock firmware that largely resolve the issue on current stock firmware, otherwise implementing my temp fix as a GUI option would probably be a good idea for the affected routers. The ability to assign DSCP to all traffic on chosen interfaces also might be a good thing to add anyways. |
|
 XycPremium join:2006-06-08 Sewell, NJ
1 recommendation | Calling it a "critical bug" on Comcast's end is probably a bit harsh. Nothing should be acting on DSCP markings unless specifically configured to, since their meaning is specific to a given network (or for links between networks by prior agreement). If you're going to act on DSCP, you should clear the markings where the packets enter the network and set them to what your QoS policy dictates (and what your command does). |
|
 | I don't think that's true, nearly every router will act on it since the WMM protocol on the wireless will almost always use DSCP at least somewhat. It depends somewhat on how aggressively the router responds though. My script clears the markings since I can't stop my WiFi from acting on DSCP. |
|
 JohkalCool CatPremium,MVM join:2002-11-13 Happy Valley kudos:6
1 recommendation | reply to jhilliar
Look what I found: »www.dd-wrt.com/phpBB2/viewtopic.php?p=773658
At least put it in Quotes. |
|
|
|
 plat2on1 join:2002-08-21 Hopewell Junction, NY | pretty sure its the same guy? seen him make posts on multiple forums |
|
 JohkalCool CatPremium,MVM join:2002-11-13 Happy Valley kudos:6 | Or a copy-cat killer.  |
|
 tshirtPremium,MVM join:2004-07-11 Snohomish, WA kudos:4 Reviews:
·Comcast
| reply to jhilliar
So if you are using said router with NON STOCK firmware, there may be a bug, that can't be controlled via the router config console for the NON-STOCK firmware? It's good to give other aheads up warning about the problem in the NON-STOCK firmware, but your title implies ComCast is at fault. many users just choose to disable WMM or if using stock firmware avoid the problem all togther
|
|
 plat2on1 join:2002-08-21 Hopewell Junction, NY | said by tshirt:So if you are using said router with NON STOCK firmware, there may be a bug, that can't be controlled via the router config console for the NON-STOCK firmware? It's good to give other aheads up warning about the problem in the NON-STOCK firmware, but your title implies ComCast is at fault. many users just choose to disable WMM or if using stock firmware avoid the problem all togther
disabling WMM is not advised for 802.11n. |
|
 tshirtPremium,MVM join:2004-07-11 Snohomish, WA kudos:4 Reviews:
·Comcast
| Guess it depends what you are doing and your wireless environment for me it's need is borderline at best on a DIR-655 out in the country with few or no overlapping networks. currently I have it off and I'm not seeing any problems from that AND eliminated one streaming glitch. |
|
 1 edit | reply to plat2on1
said by plat2on1:pretty sure its the same guy? seen him make posts on multiple forums Yeah, that's also one of my posts. I posted to Comcast's forum about this and their techs more or less confirmed the issue and that it should not have been happening....then disappeared without a trace, just trying to made sure this info is out there.
said by tshirt:So if you are using said router with NON STOCK firmware, there may be a bug, that can't be controlled via the router config console for the NON-STOCK firmware? It's good to give other aheads up warning about the problem in the NON-STOCK firmware, but your title implies ComCast is at fault. many users just choose to disable WMM or if using stock firmware avoid the problem all togther
Has nothing to do with stock/non-stock, its all in the WiFi drivers, which are the same(unless fixed by manufacturer). Yeah disabling WMM works, except you can't disable WMM on wireless N devices and still have wireless N(it reverts to G).....so, not a reasonable workaround whatsoever. It really is Comcast's fault however certain routers handle the problem better than others. Even on routers where the issue isn't noticeable on speedtests it can cause issues when doing local transfers as the internet priority would be below anything else since WMM reads DSCP.
said by plat2on1:disabling WMM is not advised for 802.11n.
More like not even possible.
said by tshirt:Guess it depends what you are doing and your wireless environment for me it's need is borderline at best on a DIR-655 out in the country with few or no overlapping networks. currently I have it off and I'm not seeing any problems from that AND eliminated one streaming glitch.
Have a package of more than 25 megabits and you pretty much have to use N. Live in a area with high-network congestion, have to use 5GHZ which also is not compatible with G. Since WMM is a mandatory spec the only real workaround is what I posted. |
|
 NetDogPremium,VIP join:2002-03-04 Parker, CO kudos:49 Reviews:
·Comcast
| said by jhilliar:I posted to Comcast's forum about this and their techs more or less confirmed the issue and that it should not have been happening....then disappeared without a trace, just trying to made sure this info is out there.
Didn't disappear.. Still looking at it.. I am wanting on the same router your using to reproduce the issue. I have seen people have issues with QoS, most of them disable QoS on their router and the problem goes away. I can't comment on something I can't reproduce or haven't seen myself.. -- Comcaster.. Network Engineer with NETO |
|
 | said by NetDog:said by jhilliar:I posted to Comcast's forum about this and their techs more or less confirmed the issue and that it should not have been happening....then disappeared without a trace, just trying to made sure this info is out there.
Didn't disappear.. Still looking at it.. I am wanting on the same router your using to reproduce the issue. I have seen people have issues with QoS, most of them disable QoS on their router and the problem goes away. I can't comment on something I can't reproduce or haven't seen myself.. Sorry, hadn't seen the reply on the Comcast Forum until just today, guess email notifications weren't setup or something. I'm mainly talking about the DSCP value in the ipv4 header itself, have you confirmed that Comcast is sending out all ipv4 DSCP with the non-default value? I think these are somewhat separate issues since there are a ton of different situations in my opinion where this could cause issues. I can readily reproduce the issues on at least 2 router models(cisco e2000 and e3000) with shibby 112 which have no problems on any other ISP, and due to my research I can pretty much guarantee you will also see the problem on the wrt610nv2 and likely the v1 as well as well as a fairly large number of earlier ones if reports are correct. Some of these may have had firmware updates to modify WMM's response to DSCP(original stock firmware will definetely show this issue), however I think it is important to have Comcast's network send out the proper default DSCP header value so that these situations are not encountered. Other routers that present this issues may be more obscure and may not receive updates. Even with the updated drivers these routers would still prioritize non-internet traffic over internet traffic, which is generally the complete opposite of what you would want. WMM's QoS just doesn't have granular control in general which leaves many of these with no proper fixes other than install in CFW and running my script. |
|
 willmo join:2013-06-03 Seattle, WA 3 edits | reply to jhilliar
Fascinating.
I can't get WAN-side captures immediately, but it does appear that on my WLAN, IPv4 packets from the Internet have DSCP 8 and 802.11e priority 1 = "background" = bad. I'm using a Netgear WNDR3700v1 with a trunk/bleeding-edge build of OpenWRT from last month (r37767).
I assume this has something to do with CDV...Comcast obviously needs to prioritize its VoIP traffic over HSI traffic on its HFC network and presumably they use DSCP to do that. Perhaps VoIP gateways (eMTAs or whatever they're called) then typically rewrite the DSCP when routing to the LAN. But I'd think HSI traffic could use "best effort" QoS rather than "background" and VoIP would still win. And it would be nice if, for HSI customers with separate router and modem, the modem would emit DSCP = 0. Maybe the CMTS+modem can't do that though.
I understand the argument that this is a bug in the router, but if it's true that the stock firmware on several popular routers doesn't rewrite the DSCP and then honors it on the wireless side, I feel Comcast should try hard to accommodate them anyway.
Anyway, I know very little about QoS, so I'll have to do more learning/investigation/testing tomorrow. |
|
 EGThe wings of lovePremium join:2006-11-18 Union, NJ kudos:9 | reply to jhilliar
said by jhilliar:More like not even possible.
FWIW, to my experience, on some devices it is indeed possible to disable it but it's not advised. |
|
 | said by EG:said by jhilliar:More like not even possible.
FWIW, to my experience, on some devices it is indeed possible to disable it but it's not advised. Possible to disable maybe...takes out wireless N with it. |
|
 EGThe wings of lovePremium join:2006-11-18 Union, NJ kudos:9 | said by jhilliar:Possible to disable maybe...takes out wireless N with it. Yep. |
|
 tshirtPremium,MVM join:2004-07-11 Snohomish, WA kudos:4 Reviews:
·Comcast
1 edit | reply to jhilliar
said by jhilliar:said by plat2on1: disabling WMM is not advised for 802.11n.
More like not even possible.
said by tshirt: Guess it depends what you are doing and your wireless environment for me it's need is borderline at best on a DIR-655 out in the country with few or no overlapping networks. currently I have it off and I'm not seeing any problems from that AND eliminated one streaming glitch.
Have a package of more than 25 megabits and you pretty much have to use N. Live in a area with high-network congestion, have to use 5GHZ which also is not compatible with G. Since WMM is a mandatory spec the only real workaround is what I posted. I guess you are right about WMM, found this early test of WMM for my router a DIR-655 that basically says that unchecking the WMM box doesn't stop WMM from functioning if N is streaming, at least in the earlier version (sent inquires to D-link). It also mentions the cisco problem. »www.smallnetbuilder.com/wireless···?start=1 Something did change however to a sony tv with built in g now functions better even when 2 ndevices are streaming, and doesn't seem to be capping our wireless beyond the actual bandwidth in use.
|
|
 | reply to EG
I absolutely can vouch for what "jhilliar" is saying and am so glad someone has the time / skill to make a fix to prove its Comcast's network. I have a cisco / linksys e3200 with ddwrt and had encountered the issue several months ago, although for me setting my wifi to wireless g was fine as I don't need the extra speed. However I was well aware of the issue and have read numerous forums where people had relatively new N routers which worked great on fios etc and then went to comcast with the same equipment, only to witness crawling download on wireless. Meanwhile Ethernet would be fine. I responded to a guy in a post several weeks ago who I think had this problem. »Slow download speeds with router
Anyways good luck, hope this gets resolved! |
|
 CheesePremium join:2003-10-26 Naples, FL kudos:1
2 recommendations | reply to jhilliar
I have an N router and never have experienced these issues. |
|
 tshirtPremium,MVM join:2004-07-11 Snohomish, WA kudos:4 Reviews:
·Comcast
| said by Cheese:I have an N router and never have experienced these issues. I believe it's more that "some brands, models" of routers are more susceptible to this fault |
|
 CheesePremium join:2003-10-26 Naples, FL kudos:1 | Good point |
|
 JohkalCool CatPremium,MVM join:2002-11-13 Happy Valley kudos:6 | reply to tshirt
Any idea what brands/models? |
|
 3 edits | - Removed post |
|
 tshirtPremium,MVM join:2004-07-11 Snohomish, WA kudos:4 Reviews:
·Comcast
| reply to Johkal
Cisco Linksys are known to have had the problem. unsure if there are more or if they found a solution, and updated since then.
Which is I think the OP's point, these users bought fairly high end (depending on when they were purchased) routers from name brand companies, and now find a significant bug, stimulated by a normal (for comcast) setting. I'd first be pissed at cisco (who seems to be dropping the ball too frequently recently) and then work on INFORMING ComCast and users rather than accusing them publicly first. |
|
 | said by tshirt:Cisco Linksys are known to have had the problem. unsure if there are more or if they found a solution, and updated since then.
Which is I think the OP's point, these users bought fairly high end (depending on when they were purchased) routers from name brand companies, and now find a significant bug, stimulated by a normal (for comcast) setting. I'd first be pissed at cisco (who seems to be dropping the ball too frequently recently) and then work on INFORMING ComCast and users rather than accusing them publicly first.
Cisco has released updates that largely fix the issue on some routers, however this setting on Comcast is simply wrong and can still cause issues(more minor) even on fixed devices, as far as I can tell not a single other ISP has done this before. It simply makes no sense that they have this classification in place going to customer networks that they don't control, I bet they are using it for their own internal network management but there is no reason they couldn't use the default one instead and simply set everything to use that value instead, otherwise they could strip the header at the network edge. Someone screwed up and Comcast just takes forever to get anything fixed that is anywhere reasonably complicated. Their ipv6 is correct even, no idea why Comcast didn't test these configuration options better though. |
|
 XycPremium join:2006-06-08 Sewell, NJ | You keep blaming Comcast for what looks to me like a Cisco bug. If you want to do QoS, you clear those bits at your edge as they come in and set them for whatever works in your QoS plan. It you don't want to bother clearing them, you ignore those bits. DSCP values only have local meaning within a network run by the same person or people. Cisco didn't clear the bits but still acted on them.
Comcast could clear the bits potentially to help people who haven't upgraded, but they aren't required to by any means. |
|
 | said by Xyc:You keep blaming Comcast for what looks to me like a Cisco bug. If you want to do QoS, you clear those bits at your edge as they come in and set them for whatever works in your QoS plan. It you don't want to bother clearing them, you ignore those bits. DSCP values only have local meaning within a network run by the same person or people. Cisco didn't clear the bits but still acted on them.
Comcast could clear the bits potentially to help people who haven't upgraded, but they aren't required to by any means.
The problem is that you can't ignore the bits, the only solution is to clear them at the gateway which is fairly technical, the WMM protocol(wireless QoS) is a mandatory spec on wireless N networks. |
|
 CheesePremium join:2003-10-26 Naples, FL kudos:1
1 recommendation | My WMM is off, I am not sure what is mandatory about it?
»en.wikipedia.org/wiki/Wireless_M···tensions - I see nothing here to indicate it's "mandatory" one bit... |
|
 | Do you still have WPA2 working now that u disabled it? I remember it also takes something important out with it... |
|
 CheesePremium join:2003-10-26 Naples, FL kudos:1
1 recommendation | It's been working, never stopped. |
|