Search:  

 
 
   All ForumsHot TopicsGallery






how-to block ads


 
Forums » US Telco Support » AT&T » AT&T Midwest » Netopia 33xx routers and high upchannel utilization
Search Topic:
Uniqs:
1392
Share Topic:
RSS topic:
toggle:
flat / full
normal / watch
Posting:
Post a:
Post a:
nonsense email messages from mailbot@yahoo.com »
« DSL drops when phone picked up / hung up  
page: 1 · 2 · 3
AuthorAll Replies


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
·AT&T Midwest

Netopia 33xx routers and high upchannel utilization

I've got a client on an Elite-S package that has a mailserver on a hosted IP. For several years, they've been running without a problem.

Starting about 6 weeks ago, when they send out to their mailing list, and the upchannel is pegged for a few hours sending to lots of sites, their Netopia 3346 started to freeze. It wouldn't respond locally, and even the DSL status lights wouldn't update... even if the phone cord was removed. It took a power reset to bring the router back.

So they replaced it with a 3347 Netopia unit, with wireless.

The problem remains. Only after a few hours of high up bandwidth, will this be a problem.

They're on a remote terminal, and I'm exploring the possibility that a change to the firmware in the RT might have brought out a problem that was lying latent in the Netopias. AT&T has already changed the pairs and cut them over to a different card in the RT, and the problem's still there. David's working on lowering the upchannel to 608 to see if that makes this lockup less likely to occur.

I'm just looking for anyone else who's seen this behavior with a current firmware version of either the Netopia 3346N-002 or the 3347-02-1006. Thanks.


koma3504
Advocate
Premium
join:2004-06-22
North Richland Hills, TX
Are you talking aout this version 7.5.1 R10-DP42

As i know i have had that happen with my Netopia Running
nta751r10_dp42-2.bin

And I am on a Rt as well.


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
7.5.1 R10-DP42 and its predecessor on the 3346, and 7.6.1 R9 on the 3347. These drops started about 6 weeks ago.


Dennis
Premium,Mod
join:2001-01-26
Algonquin, IL
·AT&T Yahoo

Host:
Chicago
Users find Hot Deals
Users find Hot Dea..
Requests for Hot D..
Home Repair & Impr..
reply to lev
Can you try replacing with a 2wire or something? I just have a hard time believing that the RT comes into play...when the symptoms seem more like your overwhelming the memory buffers/routing table in the CPE.
--
My Blog. Because I desperately need the acknowledgement of others.

Visit the Judd Family website to see my kids!


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
·AT&T Midwest

I can try it, but I'm going to wait a bit, since it just moved to a maintenance profile of 4512 down and 608 up. I'd like to see if it fails at this speed with the current CPE before changing anything.

Here are the stats after 16 hours.



reddog6102
Premium
join:2006-02-04
Jeffersonville, IN

reply to lev
I am a legacy bellsouth tech but we carry netopia business routers as well and this seems interesting.A couple questions for ya:.what type of problems are you having,what type of dslam?We have had our share of firmware woes with this router over the years.our latest problem was a wholesale mode (picture of house on interface page)that wouldnt open multiple static ip's to be public.Some other problems with other firmware was no 2+ platform compatability and random loss of sync.our current model is 3347-02-1006L/f/w761r9.What speed are you paying for?Your footprint looks like a 6meg profile according to your noise margin amoung other thing but you arent syncing at full bandwidth.you are at fast path wich means no noise profile and 4512kbs at 16db downstream doesnt seem right.For one you have to be close 99-100%capacity(cant tell from that footprint)and that can cause surfing issues.Most alcatel dslams sync at 8128k max.Why you are syncing at 4500k with a noise margin of 16 is strange to me.your noise margin should be lower.I'll wait for other techs to chime in on this one.


RadioDoc
Sortofadog
Premium,ExMod 2000-03
join:2000-05-11
Chicago, IL
·AT&T Midwest

reply to lev
Had multiple occasions of multiple different units (Westell, 2-Wire, Efficient/Siemens of various eras) do the same damn thing at home awhile back. Turned out to be a bad RT card. Since that's been replaced with a new one (and actually replaced, not just swapped with an equally bad card) I have had zero issues.

Coincidence? Nope. Not according to the last tech who was out and got it fixed for good.


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
·AT&T Midwest

reply to reddog6102
reddog, you missed some information in the first post.

The customer has elite-S service and was syncing at 6016/768, and tests about 130 feet away from the RT, for the simple reason the RT is in the basement, and they're on the 9th floor. Note the attenuation of 3.0 and 1.5. They are WELL below the capacity threshold.

This problem started 6 weeks ago, and the 3346 was operating from July 2005 until then. The problem persists with a new 3347, and only happens during times when the upchannel is being stressed by their server sending out a large mailman-based mailing list.

I asked David See Profile to set them to a profile that only had a 608 up channel, and this was the best he could find that would do that.

Radiodoc, I had your situation in mind when I was working with the tech. He moved the pair to a completely different card. Perhaps it was another bad card as well, but it's awfully hard to prove.

My road to proof of the problem is to first see if the lower up sync makes the problem go away.

If it does, I'll complain about that, and see if it coincides with an update to the RTs.

If it doesn't, I'll swap in a 2Wire OfficePortal 1800.

If that fails, then I can point back at the RT card. But no matter what, I'm collecting all data from other people in a similar situation.


RadioDoc
Sortofadog
Premium,ExMod 2000-03
join:2000-05-11
Chicago, IL
·AT&T Midwest

He tried three different ADLU cards before he got one which actually worked. Evidently there was/is some kind or austerity program going on and they just keep swapping cards without knowing the exact functional status of any of them. He finally managed to pry a "new" card out of the garage and installed it. No problems at all since.

Now, my brother in law, who is on the same RT, is having all sorts of problems with both DSL and POTS. The wiring in his building has been eliminated as the cause, and three tech visits later his DSL is down again.

I suspect most of the cards are in fact bad.


gdm
Premium,MVM
join:2001-06-15
Mchenry, IL
clubs:
I had this problem when I was in Bartlett over 3 years ago. My card went bad twice, the first time it was just DSL. The 2nd time it was POTS. The tech that came out the 2nd time said the same thing as yours Doc.


koma3504
Advocate
Premium
join:2004-06-22
North Richland Hills, TX

edit:
September 3rd, @05:59PM

reply to RadioDoc
That is some very usefull information there. Thanks for posting.

That Information would make a very good site faq.


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
reply to lev
Okay, the new profile didn't fix it.

I'm going in today with another router, but also asking David to see about getting a completely new card installed.


RadioDoc
Sortofadog
Premium,ExMod 2000-03
join:2000-05-11
Chicago, IL
While you're waiting, keep an eye on the upstream CRC errors. If they start piling up, with no corresponding increase in downstream CRC errors, you may have exactly the problem I had...which was definitely at the RT.


reddog6102
Premium
join:2006-02-04
Jeffersonville, IN

reply to lev
This sounds like a Catina/Cienna dslam(correct?).If so the new cienna cards have pinwheels on them that need to coincide with the cable pair of the shelf.Does the sync lite go out when problem occurs?Has the DLC crew checked out the common card for the shelf?Is it OC3 fed or t1 fed?Probably fiber fed if they offer 6meg.My gut tells me its the netopia router,they are very sensitive to junk trafic and will shut down/lock up.Also maybe a t1 working in the same binder group of house cable.This will suck out a certain frequency(forget wich one).


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
·AT&T Midwest


edit:
September 4th, @01:09PM

reply to RadioDoc
It's been a few hours since the last restart, and there's still only 1 upchannel CRC error.

When the router crashes, I can't get into it anymore to find the value. This ain't looking easy to solve. I'll have to take them to Covad or a T1 if this doesn't improve soon.


---

Update:

The router crashed completely about 2 minutes after I posted the above. It was not pingable on the LAN, but it did know when a cable had been removed from the ethernet port.

The DSL light went out when I unplugged the phone line, but from experience in the past, it does not establish a PPPoE session when reconnected.

I do think the upstream output power on the Netopia is higher than it used to be in this location (notice it's got a negative value for the downstream power).


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
·AT&T Midwest

I'm going to paste part of the "all-info" diagnostic of the modem in.

192.168.254.019 is the server that's sending out to a mailman mailing list.

Does anyone think the Netopia's table is overloading? There's a lot of calls for DNS on UDP 53.

The Netopia is fine on memory, as you can see first.
--------
Memory status:
Heap: total bytes 7555072 (free 4663176, allocated 2891896)
System Packet Buffers: total 500 free 468 min 346
Image usage: text 3922636, data 726404, bss 636812



Dennis
Premium,Mod
join:2001-01-26
Algonquin, IL
wow....is that server on 192.168.254.19 running unix/bind and trying to be a DNS root server or something crazy?


lev
Premium,Ex-mod 2002-08
join:2001-05-30
Chicago, IL
clubs:
·AT&T Midwest


edit:
September 4th, @02:05PM

Yes, it's running Linux, and no, it's not trying to be a root server. It's just sending out to a mailing list, and doing its own lookups for each site.

Also, it's rejecting connections from incoming mail servers that are open relays at a pretty startling rate... approximately 110,000 blocked connection attempts per week.


Dennis
Premium,Mod
join:2001-01-26
Algonquin, IL
It also seems to be running a webserver....


reddog6102
Premium
join:2006-02-04
Jeffersonville, IN

reply to lev
I was thinking the power output on the upstream looked very high myself being so close to the dslam.What i would look at is the house cable from the dslam to the router.Is it old,bridgetap,various t1's working close by.As far as the ip route you posted,thats beyond my paygrade/scope of work!
-
Forums » US Telco Support » AT&T » AT&T Midwestnonsense email messages from mailbot@yahoo.com »
« DSL drops when phone picked up / hung up  
page: 1 · 2 · 3


Tuesday, 02-Dec 08:46:40 Terms of Use | Privacy Policy | Hosting by www.nac.net - DSL,Hosting & Co-lo | feedback | contact
over 9 years online! © 1999-2008 dslreports.com.republican-creole
page compression OFF
Most commented news this week
· [99] AT&T Metered Billing Trial Hits Second Market
· [71] UDP BitTorrent Will Destroy The Interwebs!
· [57] Comcast Tries To Slow Verizon's Philly Entry
· [17] FCC To Vote On Free National Wireless Broadband
· [14] Clearwire May Slow WiMax Build
· [8] Embarq Rejected Higher Offer
· [8] Hawaii Telecom Files For Bankruptcy
· [6] Monday Evening Links
· [0] Tuesday Morning Links
· [0] EFF Challenges Telecom Immunity
Most people now reading
· Is this a good thing for the net? [news,99366]
· Level 80 PVP gear info? [World of Warcraft]
· Upverting DVD players vs Blue ray DVD players. [General Questions]
· 80 done, Naxx cleared.....can you say WOW...GG? [World of Warcraft]
· Coalition Government Possible? [TekSavvy]
· Notice, new uTorrent Alpha may be able to evade throttling [TekSavvy]
· Java SE Runtime Environment (JRE) 6 Update 11 [Security]
· Upgrade from 768kbps/128kbps to 1Mbps/386kbps? [Verizon Online DSL]