how-to block ads
|
webmaker
join:2004-05-08
San Francisco, CA
| reply to webmaker
Re: [netopia] Netopia 3346N is crashing today
I don't understand why another system, running Azureus through a Netopia, never crashes. And this system was great until Monday, when it was crashing or locking up the router all the time.
I am wondering if the Netopia needs to have some ports forwarded or something.
Thanks, | |
tschmidt
Premium,MVM
join:2000-11-12
Milford, NH
 ·Verizon Online DSL
| said by webmaker  :I don't understand why another system, running Azureus through a Netopia, never crashes. And this system was great until Monday, when it was crashing or locking up the router all the time. Didn't Netopia have problems with underspec'ed wall warts in the past? Could your problem be low AC line voltage?
Sounds like you have multiple units - try swapping power modules and see what happens.
/tom | |
webmaker
join:2004-05-08
San Francisco, CA
| said by tschmidt :Didn't Netopia have problems with underspec'ed wall warts in the past? Could your problem be low AC line voltage? Sounds like you have multiple units - try swapping power modules and see what happens. /tom
Tom, I have a Netopia 3346 at home with lots of computers running Azureus with no problems. And at work a Netopia 3346 and no problems running Azureus,, until Monday.
I read about the power supplies not being big enough, but I have never had a problem with the small power supply that came with these units. There was a big discussion here about 1250watt before, and now getting 450watt power supplies. I have the 450w power supply.
You are correct, I could switch back to my Speedstream and Linksys router at home and take the Netopia power supply to work and see if swapping them has any effect.
I just had a thought: I have a Netopia 3546-002 router sitting here at home doing nothing. I think I will take that to work and remove the 3346 and install the 3546 and see if I still have the crash problem.
Thanks all! | |
ProtocolOH
join:2008-04-19
Beachwood, OH
| said by webmaker :Tom, I have a Netopia 3346 at home with lots of computers running Azureus with no problems. And at work a Netopia 3346 and no problems running Azureus, until Monday. I read about the power supplies not being big enough, but I have never had a problem with the small power supply that came with these units. There was a big discussion here about 1250watt before, and now getting 450watt power supplies. I have the 450w power supply. You are correct, I could switch back to my Speedstream and Linksys router at home and take the Netopia power supply to work and see if swapping them has any effect. I just had a thought: I have a Netopia 3546-002 router sitting here at home doing nothing. I think I will take that to work and remove the 3346 and install the 3546 and see if I still have the crash problem. I would be very curious to hear what your resolution to this problem was. I have had the exact same problem with my 3346N-002 (SBC/AT&T) for over a year now. Azureus crashes the router within hours or days; if I shut down Azureus, it never crashes, ever. 100% reproducible.
I bought my 3346N from the AT&T shop when my old Speedstream 5861 crapped out (took it apart, found that the power supply had aged, capacitors were out of spec). I ran it for nearly 30 days, and the damn thing kept crashing. So I exchanged the 3346N for another identical unit, thinking maybe I had gotten a bum unit. No dice, second one did exactly the same thing. OK, so I exchanged that one for a third unit. Does the same damn thing, and that's the one I've got on my wall right now.
I wondered if possibly the hardware was overheating under stress. So I took the lid off (yeah, at this point, I'm figuring it's mine now, no more warranty), and used a RadioShack IR thermometer to measure chip temperatures (great little gadget). The main CPU was getting upwards of 80-90C, with a few other chips getting almost that hot... and that's with the lid off.
So I did a case mod (yeah... to my DSL router... ubergeek). Cut an opening for an 80mm fan in the top, cut two rectangular slots, one down each side of the router for air intake, and mounted an 80mm exhaust fan and steel guard to the lid. Then I applied some self-adhesive video card RAM sinks to the various warm chips, and soldered wire leads from a 3-pin header to the incoming 12V power, for the fan. Packaged it all back up, flipped it on, runs great... except when Azureus is running (dammit).
So it wasn't the router hardware overheating (I can now guarantee I have the coolest-running 3346N south of Canada). I was concerned that perhaps it was the power supply (I think the brick they supply is specced to 500mA or somesuch), so I'm running the unit off a Netgear 12V brick that gives 1.0A, with no changes. So it's not the wall wart. I suppose I could run the unit off my DC bench supply, to see how much juice it's actually drawing, but I don't think that's the problem.
Bottom line, running Azureus crashes the router reliably after a few hours or days (cycling power always restores operation, for a few hours or days at a time), and keeping Azureus off keeps the router running 100% of the time. There's no question it's Azureus inducing the failure. Since I'm an embedded systems engineer, I actually created a little black box with lights and buttons that will automatically interrupt and restore power to the router every 8, 12, or 24 hours, for reliability's sake in case I'm away from home for a day or more and the router flips out. It sucks that I had to engineer something to keep my router running reliably, but there you go.
My setup here... I've got a Linux server which also acts as a firewall to the rest of my network, which currently consists of a WiFi router and a Windows desktop machine (on which runs Azureus, when I choose to run it). The router's running firmware 7.5.1 R10 (it shipped with R2, both versions showed the problem). I have a /29 subnet (8 addresses), and I'm using the static NAT IP mapping feature to map my public addresses onto my internal private ones, including for my server. (It's the exact same address scheme that I used for my SS 5861... theoretically, I could get a +12/-12/+5 supply and resurrect my 5861, but I can't be bothered, and the 3346N is a much nicer router, apart from crashing all the damned time).
So yeah, OP, I can absolutely confirm the problem that you're seeing, and I've had it for ages. I emailed Motorola through their web form yesterday, I'll let you know if I hear anything back. | |
tschmidt
Premium,MVM
join:2000-11-12
Milford, NH
·Verizon Online DSL
| My understanding is the the wall wart was somewhat undersized and under low line condition would cause router to crash. I personally have not experienced the problem.
Have you identified when Azureus crashes the router?
My first guess would be line quality problems severe enough so the extra computer load is enough to dump the Netopia. If you have access to AC power quality measurement gear that will tell you of Mains power is the problem.
Another quick and dirty test if you have not already done so it to move the offending PC off the Netopia circuit (ideally on the other AC leg) and see if that has any effect on crashes. If it does you have pretty much isolated mains power as the root cause.
If router still crashes it may be a data problem. Are you using the Azureus box for P2P? P2P traffic is very demanding of NAT routers and can quickly fill up NAT translation table RAM.
/tom | |
ProtocolOH
join:2008-04-19
Beachwood, OH
| said by tschmidt :My understanding is the the wall wart was somewhat undersized and under low line condition would cause router to crash. I personally have not experienced the problem. I've actually replaced the original wall wart with a unit that provides 1000mA, considerably more than what the original P/S provided. No effect.
Have you identified when Azureus crashes the router? You mean the precise moment when the router takes a dive? No, it's pretty random. It always seems to run for at least a few hours (i.e. it doesn't crash 30 seconds after a restart), and the upper limit seems to be about 2-3 days.
My first guess would be line quality problems severe enough so the extra computer load is enough to dump the Netopia. If you have access to AC power quality measurement gear that will tell you of Mains power is the problem.
Another quick and dirty test if you have not already done so it to move the offending PC off the Netopia circuit (ideally on the other AC leg) and see if that has any effect on crashes. If it does you have pretty much isolated mains power as the root cause. I'll be honest... I'm rather skeptical of the bad AC line quality theory. If this were a full-fledged computer that was flaking out, I could see AC mains having a more direct effect on the operation of the system, but the router is running on an oversized AC-to-DC wall wart.
I've got two ways to vet the theory: 1) I can run the wall wart directly off my UPS... if there's an AC line glitch, the UPS will cut in. 2) I can run the router from my regulated DC bench supply, which can pump out 3 amps clean (and I can run the supply off my UPS to boot). Advantage there is I can observe exactly how much juice the router is drawing at any given time, and see if the current draw rises under high network load.
If router still crashes it may be a data problem. Are you using the Azureus box for P2P? P2P traffic is very demanding of NAT routers and can quickly fill up NAT translation table RAM. Um, Azureus is a P2P client... it's an implementation (a rather popular one) of BitTorrent. So yes, it goes without saying that this is P2P traffic. If the router is crapping out under load, that's a firmware problem that I would expect Motorola/Netopia to fix. For what it's worth, my years-old Linux box acts as a firewall for my Windows/Azureus machine, and has no trouble handling the load. I'm hoping to hear back from Motorola to see if they think this is an issue. | |
davidg
Good Bye My Friend
Premium,MVM
join:2002-06-15
Greenville, MS
clubs:
| are you uploading a lot at the tiem it craps out, or downloading, or both? if you max out the upload, then you leave no room for handshakes and the router will crash. if you have a 256k upload for example, you should limit your upload speed inthe program to no more than about 180-190k. you need at least 10% for general overhead and handshakes, and a few extra k for a buffer never hurts.
--
Lack of Preparation on YOUR Part does NOT Constitute an Emergency on Mine! | |
ProtocolOH
join:2008-04-19
Beachwood, OH
edit:
April 21st, @11:10AM
| Heh, I'd thought of that, and yes, I clip Azureus' bandwidth. Full-bore, I get about 160KB/s down and about 36KB/s up, so I clip Azureus' aggregate bandwidth to 100KB/s down and 25KB/s up. That lets me browse competently while Azureus does its thing. But no, most of the time I'm just uploading, because all my downstream torrents have completed.
Still, even if I were maxing out my bandwidth, it's no excuse for a router crashing and not recovering itself... whatever happened to a watchdog timer? (I'm an embedded systems engineer, so this type of failure offends me.) Haven't heard from Motorola yet, I hope to in the next couple of days, if their "thank you for mailing us" auto-reply is to be believed.
EDIT: Just got my response from Motorola...
"Thanks for contacting Motorola Broadband Solutions Inc. Please contact ATT for line troubleshooting and they can conference you in with us for Cayman troubleshooting."
Figures. | |
davidg
Good Bye My Friend
Premium,MVM
join:2002-06-15
Greenville, MS
clubs:
| the batwings ain't gonna do a thing to work with you on it.
the 3xxx series has always had problems with continuous uploads. some flake and lock, others never exhibit any symptoms. my 3547 does it form time to time when we send out a lot of data from work. i ran the tweak tester fromteh tools section of this site, then adjusted my machines accordingly and so far no more issues. seems that if you are getting too many fragmented packets the buffer in the router fills and never empties.
as far as watchdog timers, i would jus tlove to see one that worked as you want it! mine usually either trip out too soon or WAY too late. certain phone switches we have sometimes take a couple minutes to reboot after a software update. the watchdog will kick in and reboot before the update is done, forcing it back to the old software!
--
Lack of Preparation on YOUR Part does NOT Constitute an Emergency on Mine! | |
ProtocolOH
join:2008-04-19
Beachwood, OH
| Yeah, I figured that Motorola likely wasn't going to lift a finger. It's sad, though... they get a well-documented bug report (well, as detailed as I can put into 1500 words in their webform), and they don't want to fix it. As an embedded software engineer myself, I'm disappointed not only that they'd release a product like that, but also not bother to repair such a defect. And it is a defect... there shouldn't be anything that I can do on this side of the network, short of sending maliciously malformed packets, that will make that router crash. (Note that all my network activity for my Windows machine first goes through my Linux firewall.)
Yeah, watchdog timers are damn near standard in modern MCUs / embedded CPUs. And I know the problem you're describing, not that specific one, but in general. A couple of jobs ago, I was asked to help with fixing the board support package for a module that another group had crafted. Well, they'd farmed out the schematic design, and the designer had implemented an external watchdog device that had to be kicked every second or less, else it would apply hard reset to the CPU. That's what it's supposed to do, right? Except he also provided no means whatsoever of disabling the watchdog device... which was a friggin' pain in the ass when you're trying to initialize registers, check flash, decompress your kernel image, and boot... thing would keep rebooting itself moments after power-up. Even better, you couldn't use an external debugger with the thing... the watchdog had no idea that you had the JTAG debugger plugged in, and it would merrily reboot the CPU after not being kicked. Dumb, dumb, dumb design.
[/rant]
Anyway, here's what I'm trying. I've got a 3-amp DC bench supply, plugged into a UPS, running at 12V... that should eliminate any issues with AC line quality or an undersized wall wart. The router's sticker says it consumes 450mA at 12V, and that's just about right; with my little auto-reboot box plus the lid fan running, the whole thing is pulling just under 500mA, the supply shows.
I've also ditched Azureus and am now running uTorrent, to see if that makes a different. (Azureus, for all its fancy features and stuff, has become rather bloated of late.) I've suspended the operation of my auto-reboot thingy, so that the router can run for days at a time without a reboot, to see how stable the router is under load from uTorrent. For what it's worth, I watched the power consumption of the router on bootup, and it probably started at around 350mA or so and eventually crept up to 450mA once it got everything running. But whether there's no traffic or a full load, the power consumption doesn't budge... so again, I don't think it's a power supply issue. (The brick I was using was rated for 1A, anyway). So, there ya go. I'll report on whether the thing keeps crashing, or whether switching to uTorrent seems to be more Netopia-friendly. | |
koma3504
Advocate
Premium
join:2004-06-22
North Richland Hills, TX | Check the milliamps and voltage on your telephone line. | |
ProtocolOH
join:2008-04-19
Beachwood, OH
| said by koma3504 :Check the milliamps and voltage on your telephone line. Could you please explain to me how this would induce my Netopia 3346N to crash only when I'm running Azureus / BitTorrent? | |
koma3504
Advocate
Premium
join:2004-06-22
North Richland Hills, TX | coencidence | |
ProtocolOH
join:2008-04-19
Beachwood, OH
| OK, update... while I have not checked the milliamps or voltage on my telephone (nor the ambient temperature, relative humidity, barometric pressure, moon phase, or sunspot activity), I did make one alteration to my system... I'm now running uTorrent instead of Azureus... they're both BitTorrent clients, and I'm running the same level of activity (same torrents, trackers, bandwidth limits, etc). And it's only been three or four days, but the 3346N has not crashed... yet. I plan on running it for another week without power-cycling, and if it runs stable, I'll consider the problem solved, and chalk it up to some weird interaction between the Netopia 3346N and Azureus specifically (and not BitTorrent clients in general).
Side note: While I've always liked Azureus' interface, I'm amazed that uTorrent provides nearly all the functionality of Azureus in a client that's a 215KB (yes, kilo-bytes) executable and occupies ~5MB of RAM, as opposed to 100MB+ for Azureus. I think I'm pretty much sold on uTorrent (especially since it doesn't seem to be crashing my router). | |
|
