Connection failing mid morning while router renegotiates

[kentronix]

I thought these line checks and renegotiations of speed/noise were meant to happen out of hours ? Every monday and friday late morning (since a firmware update a few weeks ago) the router connects up to the bt servers, has a chat about errors and such and makes some changes. Resulting in loss of internet and loss of our telephone (voip through the router). It happens other times too but without fail it happens monday late mornings.

We dont want the fastest broadband we can possibly get. We want a reliable service that wont drop in the middle of the day through BT's choice. Is there a way to reduce the speed and just fix it at a stable level, or stop it renegotiating or perhaps force it to happen out of hours  ? Is there some config error somewhere causing it to happen at our busiest time of the week ?

I am guessing at some point it will find a stable level on its own but I just want to set it nice and slow but reliable.

I appreciate any advice here.

Ideallly I would like full control of our router but short of a firmware hack thats pretty much out of the question.

[MHC]

Are you sure?    It it possible that it is loosing sync becuse of a local "noise event".     I have come across a location where an adjoining factory would start a machine late morning and it was enough to cause a loss of ADSL sync.

Get a copy of Router Stats Lite (FREE) and run that with frequent update around the suspect times logging both SNR and sync speed then post an image here.

[kentronix]

Thanks for the reply, I will try that program and report back.

A chunk of my log from this morning in the router logs shows :-

INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.TransmitBlocks' value changed: '1100052' -> '236430'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.UpstreamPower' value changed: '124' -> '104'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.LinkRetrain' value changed: '2' -> '0'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.X_00D09E_LossOfSignalFailures' value changed: '4' -> '0'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.DownstreamPower' value changed: '183' -> '184'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANCommonInterfaceConfig.TotalBytesSent' value changed: '692015011' -> '706534891'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.UpstreamNoiseMargin' value changed: '78' -> '73'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANCommonInterfaceConfig.TotalBytesReceived' value changed: '2001500661' -> '2037918619'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.CellDelin' value changed: '1220' -> '77'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.ATUCFECErrors' value changed: '5014' -> '367'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.ATUCCRCErrors' value changed: '678' -> '19'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.DownstreamNoiseMargin' value changed: '28' -> '33'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.DeviceInfo.UpTime' value changed: '604411' -> '609313'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.InitErrors' value changed: '2' -> '0'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.FECErrors' value changed: '218329' -> '4646'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.CRCErrors' value changed: '2512' -> '109'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.X_00D09E_UnavailableSeconds' value changed: '83' -> '0'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.LossOfFraming' value changed: '4' -> '0'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.SeverelyErroredSecs' value changed: '48' -> '3'
INF	2012-03-12T12:35:13Z	acs	'InternetGatewayDevice.WANDevice.1.WANDSLInterfaceConfig.Stats.CurrentDay.ATUCHECErrors' value changed: '3149' -> '35'

Which looks to me like my router config getting altered.

Everything was fine until a few weeks ago when mid morning on a monday a firmware update was pushed to the router, causing a reboot and chaos around the office 🙂  

Since then its been significatntly faster and significantly less reliable. Every time we have a problem I go to the logs and find similar to above.

[nikkil]

"the router connects up to the bt servers" I think there's a work around for this. How about if you block the IP address for now so it won't establish a session regularly. 

[MHC]

Those changes occur continually ... every time data is sent or received or an error detected or teh noise floor moves ... a register will be updated.

I wold be very surprised if teh firmware was delivered mid-morning - all thise I have seen occur between 22:00 and midnight and when it happened to my router, all I noticed was a slowdown and no loss of service.

[kentronix]

Thanks for your replies and sorry for the delay/bringing back from the dead but I am still having problems with this.

I dont like running third party apoplications on the network unmonitored so I try to run the app, and watch it,  when I suspect router issues. I havent captured anything significant in terms of fluctuations because when they happen I am running around bringing our backup internet system up (3g dongle).

The current problem seems to be the router is negotiating a very high line speed with very low noise margin. Meaning any fluctuations and the line gets dropped. Is there a way to cap the speed, force a minimum noise margin or something like that ?

Its currently running :-

Down, Up
Rate: 18829 kbs, 1160 kbs
Max Rate: 18725 kbs, 1176 kbs
Noise Margin: 2.8 dB, 6.8 dB
Attenuation: 17.7 dB, 6.9 dB
Output Power: 18.4 dBm, 11.6 dBm

Now that may seem like a great connection speed but we dont need it, we need reliability. I am doing what I can to improve the reliability of the physical cabling and filter but really I just want a higher margin for error. 2.8 seems extremely low to me. Any help appreciated. Would an alternative router allow me to force these things ?