Wifi has gotten more and more unstable.
|
| Author | Content |
|---|---|
| Bob_Robertson Apr 20, 2010 7:26 PM EDT |
I know the standard reply to any problem is "latest kernel?", and I'm running 2.6.26, but let's see if there is an answer to this one. Wifi has been getting progressively less reliable. [55386.658035] wlan0: RX deauthentication from 00:0c:41:fb:d5:50 (reason=7) [55386.658035] wlan0: deauthenticated [55388.414898] wlan0: authenticate with AP 00:0c:41:fb:d5:50 [55388.416410] wlan0: RX authentication from 00:0c:41:fb:d5:50 (alg=0 transaction=2 status=0) [55388.416418] wlan0: authenticated [55388.416426] wlan0: associate with AP 00:0c:41:fb:d5:50 [55388.418779] wlan0: RX ReassocResp from 00:0c:41:fb:d5:50 (capab=0x411 status=0 aid=3) [55388.418787] wlan0: associated [55388.418801] wlan0: switched to short barker preamble (BSSID=00:0c:41:fb:d5:50) [55414.541270] wlan0: switched to long barker preamble (BSSID=00:0c:41:fb:d5:50) [55431.570718] wlan0: switched to short barker preamble (BSSID=00:0c:41:fb:d5:50) [55499.585774] wlan0: switched to long barker preamble (BSSID=00:0c:41:fb:d5:50) [55513.167518] wlan0: switched to short barker preamble (BSSID=00:0c:41:fb:d5:50) [55549.478526] wlan0: RX deauthentication from 00:0c:41:fb:d5:50 (reason=7) [55549.478526] wlan0: deauthenticated [55550.770540] wlan0: authenticate with AP 00:0c:41:fb:d5:50 [55550.775501] wlan0: RX authentication from 00:0c:41:fb:d5:50 (alg=0 transaction=2 status=0) wlan0 IEEE 802.11 ESSID:"..." Mode:Managed Frequency:2.417 GHz Access Point: 00:0C:41:FB:D5:50 Bit Rate=36 Mb/s Tx-Power=3 dBm Retry min limit:7 RTS thr:off Fragment thr=2352 B Encryption key:... Link Quality=90/100 Signal level=-58 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:0 Invalid misc:0 Missed beacon:0 then wlan0 IEEE 802.11 ESSID:"..." Mode:Managed Frequency:2.417 GHz Access Point: 00:0C:41:FB:D5:50 Bit Rate=18 Mb/s Tx-Power=3 dBm Retry min limit:7 RTS thr:off Fragment thr=2352 B Encryption key:... Link Quality=89/100 Signal level=-62 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:0 Invalid misc:0 Missed beacon:0 the bitrate is never stable for long. 1 to 54M, seeming randomly. There are no other nearby sites on this frequency, and as you can see the "link quality" is spectacular. I tried pegging the link speed to 54M and 11M, but both times it would "deauthenticate" and "reauthenticate" even more. Sometimes the kernel log is filled with the "short" and "long" preamble messages. I did a search, and there seemed to be a problem two years ago, but there were no resolutions posted. Could it be that everyone else updated their kernels? |
| montezuma Apr 20, 2010 7:38 PM EDT |
Which driver Bob? |
| Bob_Robertson Apr 20, 2010 7:54 PM EDT |
Ok, here's everything I can think of, which one is the "driver" you want? 03:05.0 Network controller: RaLink RT2561/RT61 802.11g PCI [ 249.477660] firmware: requesting rt2561s.bin r8169 27716 0 # lsmod | grep rt2 rt2x00pci 7648 1 rt61pci rt2x00lib 22432 2 rt61pci,rt2x00pci firmware_class 6816 1 rt2x00lib rfkill 5652 1 rt2x00lib led_class 3908 1 rt2x00lib input_polldev 3752 1 rt2x00lib mac80211 139808 2 rt2x00pci,rt2x00lib cfg80211 21608 2 rt2x00lib,mac80211 |
| jdixon Apr 20, 2010 9:08 PM EDT |
Bob, stupid question, but is it only your access point which is giving you problems, and if so have you tried rebooting it? The problem might not be on your end. |
| montezuma Apr 20, 2010 9:35 PM EDT |
This may be relevant: http://ubuntuforums.org/showthread.php?t=1232749 They solved the problem by dumping dhcp and using a static address. My guess is that the rt61pci driver is buggy and the workaround above avoids a bug. |
| Bob_Robertson Apr 20, 2010 9:42 PM EDT |
JD, not stupid. Actually, I tried three different routers, and an external access point. All three had the same kinds of problems, but it got steadily worse. I changed to an external antenna on a cable, which I put way up in case it was interference with my wife's PC which is about 4 feet from mine. While that seemed to work for a while, well, that was having the problems earlier today. So I went back to the antenna directly attached to the card. I'm seeing less problems, but I also saw less problems when I started with the cabled antenna too. I will leave it like this for a while and see what happens after a few days. Anyway, thanks for the link. I'll see what they say, although I must admit that googling for "switched to short barker preamble" was an interesting exercise in"several other people have seen the same problem, and no answers." |
| Bob_Robertson Apr 20, 2010 9:46 PM EDT |
Monte, I'm already using static addressing. What I find _very_ interesting is beltts05 saying that things were much better when he made a change, and then started getting bad again after a week. That's a comparable time frame for the kind of "progressively worse" problem I'm seeing. Ain't computers amazing? |
| HoTMetaL Apr 20, 2010 10:07 PM EDT |
Bob, A ton of work on the rt2x00 drivers went into the kernel after 2.6.26, including many improvements specifically for RT61-based chipsets. I literally pulled my hair out trying to diagnose a client's RT61 problems until I finally decided to install a more recent kernel, and then everything just worked (and kept working). I realize you expected this kind of answer, but for this specific wireless driver, a newer kernel could make a world of difference. It did in my own personal experience. Try 2.6.29 or better. |
| dinotrac Apr 20, 2010 11:19 PM EDT |
Bob_R - Glad I no longer feel your pain, but am happy (in a strange sort of way) to hear that you are feeling it. I used to run rt61pci and I used to think I was insane. Fortunately, I needed more better steadier bandwidth and upgraded to a Class N router and card, so I no longer need the rt61pci. It is nice to know, however, that, while I may actually be insane, rt61pci problems might not have been a manifestation of that insanity. |
| gus3 Apr 21, 2010 8:45 AM EDT |
I'm also seeing a degradation in stability. I'm willing to attribute it to the Northern Hemisphere's warmer weather and increased (real, not relative) humidity. |
| jdixon Apr 21, 2010 9:17 AM EDT |
> All three had the same kinds of problems, but it got steadily worse. Then yes, it sounds like a driver issue. :( And the solution is probably either a new kernel or to try backporting the newer module to your existing kernel (I've looked into that before. It's usually easier to install the newer kernel). Since it's a pci, it sounds like this is a built-in chipset, so replacing it obviously isn't an option. |
| techiem2 Apr 21, 2010 12:43 PM EDT |
If it's a laptop using an internal wnic, often they are a minipci card, which you could easily replace (I tend to pull them out of dead laptops I acquire. hehe). |
| Steven_Rosenber Apr 21, 2010 12:46 PM EDT |
I ran the rt73 driver in Debian Lenny - 2.6.26 - and had no trouble with it. I'm not all that confident that a newer kernel will solve your problems, but grabbing one from Debian Backports is definitely something you should try. My hunch is that there's a hardware problem of some kind going on. Either something's loose connection-wise or slowly dying. All my hardware is so damn old, I use either USB or Cardbus for all wireless networking, and my el-cheapo CNet CWD-854 USB Wifi adapter ran great in Debian Lenny, FreeBSD, OpenBSD and Ubuntus 8.04-10.04. It's cheap: [url=http://www.google.com/products?q=CNET cwd-854&oe=utf-8&rls=org.mozilla:en-US:official&client=firefox-a&um=1&ie=UTF-8&sa=N&hl=en&tab=wf]http://www.google.com/products?q=CNET cwd-854&oe=utf-8&rls=o...[/url] http://www.amazon.com/CNET-CNTCWD-854-Wireless-G-USB-Adapter... If I can solve a problem for $15-$25, I usually go that route. I have one of these and am thinking of getting a couple more. |
| Bob_Robertson Apr 21, 2010 1:10 PM EDT |
Very interesting. Thanks for the advice, and I'm glad to know that there has been a bunch of work on the drivers. I will indeed update through Backports, which I already have in the repository list, and post the results. I have a laptop, and updated the kernel there to (I think) .32, which had an external "firmware" file for the various wifi cards. Hopefully there won't be some kind of conflict when I put the new kernel in and keep .26 in case I have to jump back to what is already "working". Funny, the only reply to my post for advice to the Debian-user list pointed back to this discussion! Hahaha, I guess UseNet isn't the only place where arguments get circular. Dino, you're not crazy. Or, let me rephrase, you're not any crazier than I am. I did get a D-Link N router, because I could terminate an IPv6 tunnel on it and re-join yet another revolution. But it will not activate N without using the later encryption than WEP, which I still use because once something works I try not to change for no good reason. Maybe this is a good reason. I'll have to look up how to activate encryption other than WEP in /etc/network/interfaces, because WICD doesn't do IPv6 yet. What? Hand editing a text config file? What is this, 1995? Steve, I have used external USB adapters, but they always seemed to get, well, HOT! Maybe I do too much YouTube watching. Gus, it's obviously caused by Anthrogenic Global Warming. Isn't everything? Oh, wait, everything bad is caused by government interference in the free market. Right. Almost slipped there. Where's my "Whip Inflation Now" button? Likely in the drawer with my old Interop badges. JD, it is actually a PCI-e card I bought separate from the mobo. Heck, I didn't even buy it from the integrator, I got it from NewEgg afterwords. When I build a system, it's a camel! Or maybe a Frankenstein, being bits from here, bits from there.... Anyway, off to dselect to see what the latest kernel from backports is, more later. What an adventure! |
| Bob_Robertson Apr 21, 2010 1:20 PM EDT |
Dag nabit! Backports is ... incomplete. linux-headers-2.6.32-bpo.4-amd64 depends on linux-kbuild-2.6.32 linux-kbuild-2.6.32 does not appear to be available So much for .32 Let's try .30... |
| tuxchick Apr 21, 2010 3:17 PM EDT |
So if the Linux kernel is truly modular, why do you need to update the whole freaking kernel to get an updated wireless driver? |
| gus3 Apr 21, 2010 4:02 PM EDT |
It's modular, like the KL-10 was modular. Only parts for the KL-10 would fit it. |
| azerthoth Apr 21, 2010 4:23 PM EDT |
Bob, instead of going backwards with kernels, why not forwards? Doing a custom debuntoid kernel isnt all that rough if you dont want to upgrade the whole thing to what ever spastic animal the latest is. |
| Bob_Robertson Apr 21, 2010 4:27 PM EDT |
Well, I'm not ready to give up on Lenny yet because of KDE3.5. And I prefer to use kernels other people build, because I'm paranoid of getting something wrong. Ok, 2.6.30-686 installed. I tried the -amd64 kernel, but for some reason the ati driver choked, and choked, and choked. Fascinating. Anyway, up and runnning, and so far so good. I have seen 48M a couple of times in iwconfig, but it seems far, far more stable at 54M than before. The signal is reading 64/70, rather than 80/100, so there are certainly some fundamental underlying changes. I'm also running a 1/2hour file transfer at full speed, ~2.0Mbps, and I'm not seeing any "short/long" preamble messages with dmesg. As before, the test will be over time. It seems to me that basic Lenny has about overstayed its welcome. Time to make some hard decisions about KDE3.5... |
| Sander_Marechal Apr 21, 2010 4:30 PM EDT |
@TC: It's modular as in, you can load and unload modules to just use what you need. It doesn't imply swapping modules with different kernel versions. |
| Sander_Marechal Apr 21, 2010 4:32 PM EDT |
Quoting:It seems to me that basic Lenny has about overstayed its welcome. Time to make some hard decisions about KDE3.5... I'm loving Squeeze. But I don't use KDE... |
| dinotrac Apr 21, 2010 4:34 PM EDT |
Sander -- Although, that is actually a compile time option...you don't have to turn on kernel version checking when compiling a kernel, but... Given the relative instability of kernel APIs, there might not be much gain. |
| Bob_Robertson Apr 21, 2010 4:41 PM EDT |
Sander, I'm getting a laptop back from HP service tomorrow, it's going to be my testbed for a honest-to-ESR AMD64 install. My first. So far I expect to use xfce, but I _like_ Kmail, Konqueror, K3b, KDM and such. So it'll be a bastard child of convenience, hopefully the KDE3 to 4 data moves are successful or I'm going to be re-entering a whole bunch of userIDs and passwords. Which reminds me, I need to see if Kwallet will print a clear-text dump of its entire contents, just to be sure. Just rechecked dmesg after 3 gig of this file transfer, and not a preamble message in it. Looks like it might very well have been 2.6.26, which would also account for how the search for other people's problems seemed to have all been in 2008, and vanished but were never fixed. TC, if the now ancient presentation by Greg Kroah Hartman on the "rate of change" of kernel internals has continued, I can understand why drivers from diverse kernels would be incompatible as is. |
| Bob_Robertson Apr 22, 2010 9:33 AM EDT |
Anyway, back to networking. I just had an event. After a morning deliberately streaming CCTV to the wife's Windows machine to load the network, I received an error: [ 5632.232084] wlan0: no probe response from AP 00:0c:41:fb:d5:50 - disassociating This is the only message in the log since bringing up the interface, so it's doing MUCH better than with 2.6.26. Still, one wonders. Link quality is 66/70, so the "no probe response" isn't a problem of signal strength. More as I see it, suggestions gladly entertained. |
| Bob_Robertson Apr 22, 2010 3:45 PM EDT |
Interesting. [22746.248041] wlan0: no probe response from AP 00:0c:41:fb:d5:50 - disassociating [23398.184748] ADDRCONF(NETDEV_UP): wlan0: link is not ready [23398.193060] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 1 [23398.393050] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 2 [23398.593040] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 3 [23398.792069] wlan0: direct probe to AP 00:0c:41:fb:d5:50 timed out [23475.838204] ADDRCONF(NETDEV_UP): wlan0: link is not ready [23475.844302] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 1 [23476.044039] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 2 [23476.244036] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 3 [23476.444043] wlan0: direct probe to AP 00:0c:41:fb:d5:50 timed out What you see is one of those "disassociating" messages, and subsequent attempts to bring the interface back up. Obviously, none worked. Rebooting didn't work. Powercycling the machine didn't work. Leaving it off for an hour, that worked. So very strange. |
| dinotrac Apr 22, 2010 3:48 PM EDT |
>Leaving it off for an hour, that worked. Heat? |
| Bob_Robertson Apr 22, 2010 5:01 PM EDT |
> Heat? That or capacitance. Either way, if it gets worse, I'm getting a new card.What was that N-card you got? |
| jdixon Apr 22, 2010 5:04 PM EDT |
> Heat? Sounds like it. You got any freeze spray, Bob? |
| gus3 Apr 22, 2010 5:10 PM EDT |
Dry ice shouldn't be difficult to come by. |
| Bob_Robertson Apr 22, 2010 5:40 PM EDT |
> freeze spray Not yet... > Dry ice I recall someone overclocking using florinert and liquid nitrogen... |
| dinotrac Apr 22, 2010 5:43 PM EDT |
Bob - I didn't get an N-card, but a USB transceiver with a nifty little cable that I hook up to the port in back of my mythbox. That way, I can have the actual transceiver sitting discreetly up by the base of my TV for optimal reception. It's nothing fancy -- just a Belkin N+ USB doodad, but it's worked very well for me. I can now handle 5 high-def TV shows at once. I think 6, but I haven't actually taken it that far. |
| gus3 Apr 22, 2010 5:47 PM EDT |
A freezer and several bottles of vodka, plus one motherboard with all jumpers removed. Have fun! |
| Sander_Marechal Apr 22, 2010 5:58 PM EDT |
@Bob: You're probably thinking of the 5 Ghz project: http://www.tomshardware.com/reviews/5-ghz-project,731.html |
| Bob_Robertson Apr 22, 2010 7:12 PM EDT |
The first time I saw (correctly spelled) Fluorinert, it was a Cray-2 at Nasa Ames. I noticed that the clear encasement had.... bubbles! And then I saw the FOUNTAIN labeled "Cray Research", See here: http://en.wikipedia.org/wiki/Cray-2 Sorry, Sander. I like what you found, but no, this is it: http://www.octools.com/index.cgi?caller=articles/submersion/... "EUREKA! I have found it. Just the fluid I needed. Of course I also want to supercool this liquid. Now how do I cool this liquid??? Since I have already gone extreme by thinking of fully submerging the system, why not go really, really extreme. What do I need to add in the picture to make it more intriguing for you guys? How does LIQUID NITROGEN sound??? Yes that's it! That's what I'll do...that will be my mission. A liquid nitrogen cooled Fluorinert™ to fully submerge the motherboard and all its components!" The guys (sorry Gals, but Guys are much more likely to push the envelope because of being more suicidal) who did this FROZE the Fluorinert. Who'd'a thunk it? |
| Bob_Robertson Apr 23, 2010 12:01 PM EDT |
Locked up again this morning. Went out to the garage and pulled out a Netgear card I for some reason put away. Now I remember why, it's "ndiswrapper only". Using a USB adapter right now, is working fine, but gets REALLY HOT with use. I'll change back to the old card ...wait, the wife's machine as a pci network card in it. Maybe this Netgear will work for her.... {sinister smirk} and she'll never notice the difference. Ahhh! I'll let you know if it works. |
| Bob_Robertson Apr 23, 2010 2:38 PM EDT |
The USB dongle worked for a couple hours, and then stopped with the same error and would not restart. It wasn't even particularly hot. Back to the Realtek, it's working for now, but I ordered a refurbished Linksys. Let's see what happens next! |
| gus3 Apr 23, 2010 4:23 PM EDT |
The same error? Then it's starting to look suspiciously like a code infrastructure problem, rather than a driver problem. |
| Bob_Robertson Apr 23, 2010 6:28 PM EDT |
Agreed. I'm thinking about swapping out my AP, since I'm using a Linksys after turning off the wireless on the D-Link router, as a test to see if the problems got better. Anyway, at least after today's machinations, I'm getting a new symptom: Reconnects: [13769.801787] wlan0: deauthenticated (Reason: 7) [13770.801051] wlan0: direct probe to AP 00:0c:41:fb:d5:50 try 1 [13770.802662] wlan0 direct probe responded [13770.802671] wlan0: authenticate with AP 00:0c:41:fb:d5:50 [13770.804763] wlan0: authenticated [13770.804771] wlan0: associate with AP 00:0c:41:fb:d5:50 [13770.806964] wlan0: RX ReassocResp from 00:0c:41:fb:d5:50 (capab=0x411 status=0 aid=3) [13770.806973] wlan0: associated Repeat several times, randomly. |
| gus3 Apr 23, 2010 7:55 PM EDT |
Oh, well, yeah, it could be the AP as well. Duh. |
| jdixon Apr 23, 2010 10:41 PM EDT |
> it could be the AP as well. I remember our introduction to electronics course I took. The first step in troubleshooting a problem was to "assume there's only one problem". I thought then it was a very unwise and possibly invalid assumption. I've unfortunately been proven correct more times than I care to count. |
| gus3 Apr 23, 2010 11:57 PM EDT |
Quoting:The first step in troubleshooting a problem was to "assume there's only one problem".It's a basic re-statement of Occam's Razor, not universal, but practical. |
| montezuma Apr 24, 2010 12:17 PM EDT |
Yeah interesting. Twice in the last five years I have had something go wrong caused by two unrelated things. I couldn't believe it both times. Lightning does indeed strike in the same place twice given enough time. Occam's razor is only true in a probabilistic sense i.e. the simplest explanation is *usually* but not always correct. |
| Bob_Robertson Apr 26, 2010 10:31 AM EDT |
Ok. Latest update, for those of you that just can't stand to not know. The Linksys WAP-54 was decommissioned, the D-Link 610 is in place. The logs of three Debian Lenny installations show the same kinds of messages: [ 8405.855112] wlan0: deauthenticated (Reason: 2) [ 8406.853034] wlan0: direct probe to AP 00:26:5a:d1:75:e2 try 1 [ 8406.855513] wlan0 direct probe responded [ 8406.855521] wlan0: authenticate with AP 00:26:5a:d1:75:e2 [ 8406.858308] wlan0: authenticated [ 8406.858316] wlan0: associate with AP 00:26:5a:d1:75:e2 [ 8406.861912] wlan0: RX ReassocResp from 00:26:5a:d1:75:e2 (capab=0x431 status=0 aid=1) [ 8406.861921] wlan0: associated While they are all running 2.6.30 today, over the weekend I also had 2.6.26 and 2.6.32 installed and saw the same kinds of messages. There are no other people in this neighborhood that run on wifi channel 2, and none close enough for interference as far as I can believe anyway. Ok, I just noticed something really strange. The disassociations seem to be happening with spreads of 180 seconds, or multiples of 180 seconds when it doesn't happen every time. That's just plain sick. Other than bringing in the FCC with one of their multi-spectrum illegal-transmitter-locator trucks, any suggestions? In the mean time, I'll start looking for some timer on the router that says "180 seconds". |
| Sander_Marechal Apr 26, 2010 11:47 AM EDT |
Just to rule out something else gone haywire on your Lenny box, could you try the Linksys or D-Link in a different PC with a clean software stack? E.g. an Ubuntu Live CD? Does the problem reoccur? Also at 180 sec intervals? |
| Steven_Rosenber Apr 26, 2010 2:08 PM EDT |
Yeah, I'd run in a live environment to do some testing. |
| Bob_Robertson Apr 26, 2010 5:27 PM EDT |
Good idea. I have two laptops here, and I think I have a PCLinuxOS sitting around here somewhere. I'll do that and get back. |
| Rob_on May 02, 2010 6:25 PM EDT |
@tuxchick Apr 21, 2010 "2:17 PM EST So if the Linux kernel is truly modular, why do you need to update the whole freaking kernel to get an updated wireless driver?" Below I use "service(s)" to mean just about anything, as kernels go. Ok ? A kernel being modular only means those "services" it runs, are not part of the kernel-core. When building a kernel from source, choices to make are, to make a "service" a module or not. There would be little need for the handfull of kernels that are available, if one kernel did fit all, and all needs. People use different kernels different ways; for a kernel provider, it is a waste of space and time to try and compile everything in, regardless of what it is because *you* think everyone may need them. Then provide the 100lb source code. |
| Rob_on May 02, 2010 6:54 PM EDT |
@Bob_Robertson I know what you are going through, been there. Here is a link, you may already know it, but it list the drivers in your kernel version. http://kernelnewbies.org/Linux_2_6_26#head-20cd551f7a74cf048... The name "kernelnewbies" means little here, it is a good source for info most of the time. Take care with "updates", some of the time, they also update other things you don't see them doing for the update you're doing. Sorry I don't have any specific info for you, this kind of problem takes a large eyeglass care and time. |
| Bob_Robertson May 02, 2010 6:58 PM EDT |
Ok, quick update. All the systems are seeing the "disassociated" kinds of errors, even when running PCLinuxOS or Knoppix. While that evening there was a specific time, that has ceased to be true. Sometimes hours, sometimes seconds. I'm doing Debian AMD64 Sid with the .32 kernel right now, but forgot the Broadcom driver, damn them all to eternal torment, so I'll have to get back on that. |
| Sander_Marechal May 03, 2010 4:54 AM EDT |
@Bob: Are those disassociated systems using the same hardware or wifi chipset? If not, then it starts to sound like there's nothing wrong with the software at all but that there is some kind of interference. |
| Bob_Robertson May 03, 2010 9:08 AM EDT |
I agree, it's looking more and more like interference of some kind. No, the chipsets are completely different. The one with the most trouble re-synching is the desktop RaLink RT2561/RT61. I have two other PCI network cards, one Linksys-N and one Netgear-G, but both have @#$%^& unsupported Broadcom chips. One laptop has Broadcomm B43xx, which is supported, and is today running the Sid kernel above until I give up trying to get everything I want to work working. I tried getting an Atheros mini-PCI-e card for it, but HP tweaked the ROM to disable any network card at boot time other than the one they ship with it. The other laptop has an Atheros AR242x. I've got all three running right now, and I'll see if the problems are showing up on both over the next few hours. Really funny thing, yesterday while I was putting Sid on the HP, I had the laptop sitting on top of the cabinet (wood) in which the router sits, 20 feet away in the next room from the system I'm seeing have so much trouble, so I could use the wired network interface for the install. While it was sitting there and powered up, the network errors were so bad that even my wife's Win7 system was losing its network enough to be a problem for a 3-year old trying to get to PBSkids.org. What's sad is that the D-Link router does not allow external antennas to be used, so I can't put the high-gain or 8-foot "put it way up" extended antenna on it. Too bad, that would be a useful test. While there are several wifi nets within sniffing range, sometimes 4 sometimes 9, hey, it's suburbia, I made sure to pick a channel that no one else was using. As you can imagine the number using channel 6 with the essid "linksys" is quite depressing. So far this morning, [ 86.022210] rt61pci 0000:03:05.0: firmware: requesting rt2561s.bin [ 86.469811] ADDRCONF(NETDEV_UP): wlan0: link is not ready [ 87.228073] wlan0: authenticate with AP 00:26:5a:d1:75:e2 [ 87.230972] wlan0: authenticated [ 87.230981] wlan0: associate with AP 00:26:5a:d1:75:e2 [ 87.233877] wlan0: RX AssocResp from 00:26:5a:d1:75:e2 (capab=0x431 status=0 aid=1) [ 87.233885] wlan0: associated [ 87.235210] ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready [ 87.327123] cfg80211: Calling CRDA for country: US [ 97.481022] wlan0: no IPv6 routers present [ 499.994491] wlan0: deauthenticated (Reason: 2) [ 500.992033] wlan0: direct probe to AP 00:26:5a:d1:75:e2 try 1 [ 500.994000] wlan0 direct probe responded [ 500.994008] wlan0: authenticate with AP 00:26:5a:d1:75:e2 [ 500.996275] wlan0: authenticated [ 500.996283] wlan0: associate with AP 00:26:5a:d1:75:e2 [ 501.000529] wlan0: RX ReassocResp from 00:26:5a:d1:75:e2 (capab=0x431 status=0 aid=1) [ 501.000538] wlan0: associated [ 1174.984950] wlan0: deauthenticated (Reason: 2) [ 1175.985038] wlan0: direct probe to AP 00:26:5a:d1:75:e2 try 1 [ 1175.987522] wlan0 direct probe responded [ 1175.987530] wlan0: authenticate with AP 00:26:5a:d1:75:e2 [ 1175.990397] wlan0: authenticated [ 1175.990405] wlan0: associate with AP 00:26:5a:d1:75:e2 [ 1175.992805] wlan0: RX ReassocResp from 00:26:5a:d1:75:e2 (capab=0x431 status=0 aid=1) [ 1175.992813] wlan0: associated I'd love to know why "no ipv6 routers present" comes up, since the reason for putting the D-Link in place was specifically to turn on IPv6. =========== $ ping6 -c 5 -n gateway6 PING gateway6(2001:470:1f07:72a::1) 56 data bytes 64 bytes from 2001:470:1f07:72a::1: icmp_seq=1 ttl=64 time=1.08 ms 64 bytes from 2001:470:1f07:72a::1: icmp_seq=2 ttl=64 time=3.17 ms 64 bytes from 2001:470:1f07:72a::1: icmp_seq=3 ttl=64 time=2.29 ms 64 bytes from 2001:470:1f07:72a::1: icmp_seq=4 ttl=64 time=1.81 ms 64 bytes from 2001:470:1f07:72a::1: icmp_seq=5 ttl=64 time=0.966 ms --- gateway6 ping statistics --- 5 packets transmitted, 5 received, 0% packet loss, time 4005ms rtt min/avg/max/mdev = 0.966/1.866/3.170/0.814 ms ========= As per suggestions on the Hurricane Electric site, I have the D-Link IPv6 autoconfiguration set for "stateful" rather than "stateless", maybe that's why. Anyway, all three have seen several disassociation/reassociation cycles since I started writing this (but then it has been an hour). So interference is looking more and more likely. |
| Sander_Marechal May 03, 2010 9:45 AM EDT |
Then I'd suggest giving the FCC a ring. |
| gus3 May 03, 2010 11:35 AM EDT |
The FCC doesn't care. They demonstrated that in spades last year, when they did nothing about the "expiring car warranty" robo-call spammers, until Senator Schumer got harassed by them. And then, it was the Federal Trade Commission that got involved. (The spammers had committed a federal crime by calling cell phones, which made it an FCC matter.) http://www.npr.org/templates/story/story.php?storyId=1039985... |
You cannot post until you login.