Internet drops w\ TimeWarner cable modem -they claim it's the wireless router

[mulevad]

Hi, everyone.

I have Time Warner Cable and they provide me with a Scientific-Atlanta Cable Modem (Model: SA DPC2203).
I am using a brand new Netgear N600 Wireless Dual Band Router (Model: WNDR3400v2).

I don't think I have a problem with the wireless connection. However, several times a day the internet becomes unavailable.
If I wait for a while, sometimes 5 mins, it rights itself and everything is fine. Other times I have to power cycle the modem and the router to restore internet access.

When the internet goes down I can ping both the router and the modem and log into both. The router shows no internet connection.
I don't really understand the modem info, but everything I look at says it is running fine. I found a "log" page on the modem that says
DOCSIS (CM) Events:
Time   ID   Level   Description
 Sat Feb 09 15:49:58 2013      73040100     Notice (6)     TLV-11 - unrecognized OID
 Sat Feb 09 15:49:41 2013      82000200     Critical (3)     No Ranging Response received - T3 time-out
 Sat Feb 09 15:49:34 2013      84000100     Critical (3)     SYNC Timing Synchronization failure - Failed to acquire QAM/Q...
 Sat Feb 09 15:49:32 2013      82000400     Critical (3)     Received Response to Broadcast Maintenance Request, But no Un...
 Sat Feb 09 12:40:27 2013      73040100     Notice (6)     TLV-11 - unrecognized OID
 Sat Feb 09 12:40:17 2013      2436694060     Critical (3)     Booting up the cable modem

and things of this nature. I haven't been able to correlate the downtime with the log messages, but I see the "T3 time-out" a lot around the time things seem to go south. "TLV-11 - unrecognized OID" happens too. But right now I have internet access.

I usually have 8 or 9 devices connected to the wireless router and I never lose the wireless connection.

Does anyone have any ideas how I can convince the cable people they have a problem on their end?

They tell me the problem is the router, and ask that I connect a PC to the modem to prove there is no problem. This is not really practical since there are so many devices using the internet in my home.

I tried running a wscript that just pings and ran that for about 4 hours. With the PC connected directly to the modem I got:
2/9/2013 4:29:05 PM Ping statistics for 74.125.225.72:
2/9/2013 4:29:05 PM     Packets: Sent = 10605, Received = 10529, Lost = 76 (0% loss),
2/9/2013 4:29:05 PM Approximate round trip times in milli-seconds:
2/9/2013 4:29:05 PM     Minimum = 31ms, Maximum = 100ms, Average = 36ms

There were 76 of these: "2/9/2013 3:50:12 PM Request timed out." in my output.

I ran again for about an hour using the router and got:
2/9/2013 5:55:40 PM Ping statistics for 74.125.225.3:
2/9/2013 5:55:40 PM     Packets: Sent = 3832, Received = 3765, Lost = 67 (1% loss),
2/9/2013 5:55:40 PM Approximate round trip times in milli-seconds:
2/9/2013 5:55:40 PM     Minimum = 33ms, Maximum = 396ms, Average = 47ms

There were 67 of these: "2/9/2013 5:51:48 PM Request timed out." in my output.

Help!!! Does anyone have any idea what I can do ? Is it the router or the modem?

Thanks,
Dave Lum

[Geek-9pm]

You wrote...

I usually have 8 or 9 devices connected to the wireless router and I never lose the wireless connection.

Does anyone have any ideas how I can convince the cable people they have a problem on their end?

First, ruining n8 devices on wireless is pushing the envelope.  Just one faulty device in the wireless system can cause a dropout for all other wireless clients.

Next , you need to ask your ISP what they consider prof of bad service. Myself, I have dropout at least once a day, but the modem recovers in less that 60 seconds. So I do not even report it.

Understandable your ISP is doing to blame your other devices. Statistically they are right. You need to run a test for at least 8 hours using only the modem. Rather that do pings, I would recommend  you try downloading large files. Or watch Netflix  or Hulu for hours at a time. If a dropout happens during a video stream, you will notice it. And it will show in the error log.

EDIT: When you have a Tcpip event, it looks like this in event viewer.

But first you must clear the event viewer, otherwise you find too much stuff. Above you see two events tat came when I cut power to my ATT modem.

[mulevad]

I'll try running something overnight, or while I'm at work and the kids are at school. I ran the ping test last night (7:23 pm to 10:13 am) and we lost internet for about 11 minutes and it recovered. There were 203 failures through the night. Only about 40 of them over 30 seconds. Only the 11 minute failure was over 1 minute.

Unfortunately, I didn't run it connected directly to the cable modem as I didn't get your suggestion until today. I'll try direct connection tonight.

How could 1 bad device connected via wireless connection to the router affect the connection from the router/modem combination to the internet?

btw: our wireless devices consist of:
 3 iPods
 1 net-book
 1 tablet
 5 cell phones + more w/ guests
 3 PCs + more w/ guests
 1 TV
 1 blue-ray device
 1 WiFi router
 1 WiFi extender
 
I don't consider that too much bandwidth for my provider. They are seldom all in use.

Also, I can filter the event log and there was only 1 event that had source in {Tcpip, TCPIP, Tcpip6, Tcpip-SQM-Provider}. It was a IP address conflict during a time when internet access was fully functional.

[Geek-9pm]

If you want your provider to listen, stop using ping testing. They are more impressed with TCP/IP errors in the log.

Just one  IP conflict indicates something is not working. You should never have a conflict.  This kind of error is not a bandwidth issues,. it is about things working right.  DHCP is a low-level thing. If it fails, some hardware is acting bad.

[mulevad]

Is there something else I should use instead of ping testing?
I don't have any TCPIP errors to impress anyone with. I just know that I lose DNS, and the router doesn't have internet access.

The IP address conflict was weeks ago. I'm losing internet several times daily.

[Geek-9pm]

I am using a brand new Netgear N600 Wireless Dual Band Router (Model: WNDR3400v2).

You have the manual, -Right? I think that router has a way of logging failures with the internet connection. If so,  take a look and see if the errors on with the internet or will local traffic. And find the error code number.
You can not use ping to verify the kind of error your have. You have to find a log from neither your Windows system or the internal log of the router. The log should have a code that enumerates the error type.

ftp://downloads.netgear.com/files/WNDR3300_SM_12Nov07.pdf

[BC_Programmer]

SYNC Timing Synchronization failure - Failed to acquire QAM/Q...

I honestly cannot figure out how this could <not> be a problem on their end. This is the modem quite literally saying it cannot connect to the ISP network. (I guess is could be the cable wiring)

The problem is you have to get to a level of support where the people actually know things, rather than just following a script. Otherwise, no matter what you say they are just going to be following the script before they give up.

[mulevad]

Ok, thanks guys.

I'll look into reviewing the manual thing to find some kind of router logging. Although the NETGEAR tech support team told me to just use ping. The guy didn't sound like a NETGEAR expert so much as a 3rd party PC support salesman.

@Mastermind - I totally agree. We've called them 4 times. They sent out a guy who came out once and replaced the modem, some of the cable connections, and some of the cable he said was old. It was actually worse after he did that. Each time we call them we end up with a connect the modem directly to a PC and watch it for 4 hours.

This may take a couple of days.

[mulevad]

I ran the first overnight with a direct connection from the PC to the Modem. There were 3 significant disconnects of 6.8, 3.8, and 1.8 minutes during which times the internet was unavailable. (The disconnects started at 11:54:56 PM, 2:45:25 AM, and 3:16:09 AM respectively.)

There were no TCPIP type events in my event log. I did identify several event logs with source "DNS Client Event", but they didn't seem to correspond to the timeouts.

The manual was a bust. There was no information about event logs for the router. I will contact NETGEAR again too, and see what they have to say.

I'm still looking for help.

 

[Computer_Commando]

What do the modem signals look like?

• Downstream Power Level
• Upstream Power Level
• Downstream Signal to Noise Ratio

The symptoms you describe can be due to one or more of these values being out of range or borderline.

[Geek-9pm]

The modem should have a diagnostic that give s the info Computer_Commando named in his post above.
What was the model of the Modem? It is a Scientific Atlantic, -- Right? But what model number.

[mulevad]

Here's some of the info from the modems web page. I got the information on how to log into it from here: http://forums.cable-modem.net/index.php?action=vthread&forum=2&topic=1001

System Tab
Name - SA DPC2203
Hardware Version - 1.1
Receive Power Level - 0.6 dBmV
Transmit Power Level - 47.5 dBmV
Cable Modem Status - Operational
Vendor - Scientific-Atlanta, Inc.
Boot Revision - 2.1.6lR7

Signal Tab
Downstream Status - Operational
Channel ID - 2
Downstream Frequency - 111000000 Hz
Modulation - 256QAM
Bit Rate - 42884296 bits/sec
Power Level - 1.1 dBmV
Signal to Noise Ratio - 38.1 dBmV

Upstream Status - Operational
Channel ID - 6
Upstream Frequency - 32000000 Hz
Modulation - 64QAM
Symbol Rate - 2560 Ksym/sec
Power Level - 47.2 dBmV

Let me know if you think the power signals are ok.

Here is also a look at part of the configuration file for the modem, according to...
Config Tab
CM Configuration File
NetworkAccessControl = 1
UpstreamServiceFlow =
ServiceFlowReference = 1
QualityofServiceParameterSetType = 7
ServiceClassName = 'us_hsd_res'
ServiceFlow SchedulingType = 2
UpstreamMaximumSustainedTrafficRate = 1024000
MaximumTrafficBurst = 6144
MaximumconcatenatedBurst = 6144
IPTypeofServiceOverwrite = 0x0000
DownstreamServiceFlow =
ServiceFlowReference = 5
QualityofServiceParameterSetType = 7
ServiceClassName = 'ds_hsd_res'
UpstreamMaximumSustainedTrafficRate = 15360000
MaximumTrafficBurst = 10000000
PrivacyEnable = 1
<there was more, but it was pretty cryptic>

I don't know how to change the settings, but if it can help anyone determine anything....

Thanks in advance

[mulevad]

 
So - I don't know why no one has any other ideas.

What is a good way to monitor an internet connection?

[mulevad]

Well, I called the Cable Company again and told them that I had bypassed my router and connected directly to the modem.
They are sending a "Supervisor" out to see what is wrong with my connection.

Thanks everyone.

btw: I think it depends a lot on who answers the call.

I'm still curious what else to use besides a ping test.
I used the windows scripting host with the following network_ping.vbs file contents:

Code: [Select]

hostIp      = wscript.arguments(0)
logfilename = wscript.arguments(1)
Set fso     = CreateObject("Scripting.FileSystemObject")
Set Shell   = CreateObject("Wscript.Shell")
' OpenTextFile Method requires a Const value
' (Over)Write = 2  Append = 8   
Set logfile = fso.OpenTextFile(logfilename, 8, True)
shellstring = "%comspec% /c ping -t " & hostIP
Set oExec   = Shell.Exec(shellstring)
wscript.echo "Ping Error log With Timestamp - Ctrl + C to halt"
Do While oExec.StdOut.AtEndOfStream <> True
      pingline = Date & " " & Time & " " & oExec.StdOut.ReadLine
      logfile.WriteLine(pingline)
      If InStr(pingline, "TTL=") = 0 Then
         logfile.WriteLine("-----------------")
      End If
Loop

I run it at the command line like this: wscript network_ping.vbs google.com fileout.txt

Then I review the txt file for disconnects. I can drop it into Excel and generate a graph of when I have disconnects.



The left hand (vertical) scale is number of ping failures. It's about 5 seconds between ping failures.


[recovering disk space, attachment deleted by admin]

[Geek-9pm]

You could do more research on this, but here is the nutshell.
Ping is used simply to confirm the network in usable at the present moment. Constant pings annoy some server administrators, so they will stop responding to t a stream of pings. In short, pings are a form of intrusive testing. Administrators don't like it. So you don't want to do 24 hours of one ping  per second o yahoo or somebody else.

There are a number of software p[products to monitor actual network performance. Normally there are a few errors from time to time. These are corrected 'on the fly' and the user may not be aware of it. The monitor software collects the error reports and builds a report with  a statistical study of failure rates and severity.  Errors easily recovered are  'soft errors' and are a lessor component of the report.

The reports must be based on real data transfers, and not just ping tests.

Of course this is not new information. But many of us just don't work full-time in network management and are not aware of what tools are used.

Here is a free program you might like to try:
http://www.softpedia.com/get/Network-Tools/Network-Monitoring/Server-Monitor.shtml

Yeah, the name is Roadkil's Server Monitor, but it is a serious program.