DSL Service is Flapping

call in and schedule a tech on demand - a tech on premise is the only way to get things fixed but then you will probably have to pay a fee of $99 for it..

There are no AT&T engineers on this forum.  There is only a small team operating under ATThelp which are not too far off the same skill level you reach when you call the 800 number.   But you do get us, the other experienced users of the AT&T Community.

Your issue might be a line issue.  Or it might a utilization issue.  Or something else.  Let's start by having you go to http://192.168.1.254/ , click on the Broadband tab, and take a screen capture of the tables including the sync rates, attenuation, noise margin and the error counts.  Get the IPv4 Statistics and IPv6 Statistics too, if you would.  Include them in replies to this topic so we can analyze them.  Then we can hopefully give you a more useful suggestion than dispatch-on-demand.

The system information tab will also list the firmware version and that’s useful as well.  

If you spent that much time on the phone with Att, they should have already diagnosed a line problem.   

Do not let them send another gateway until basic troubleshooting is done.  

Also not clear what was done to get things working again.  Was it just a reboot?   

Thanks for the responses!  All helpful insights.

@JefferMC -- Here are those screenshots you requested from the Broadband Tab on the ARRIS...

Okay, first thing to note is that your lines are in great shape.  You're provisioned for 34 Mbps down / 7 Mbps (which fits very well with a nominal service of 25 Mbps) with dual pair service.  There's a good bit of spare Sync rate above that provisioning, your Noise ratio is fine.  Your attenuation is somewhat high, that really only means your lines are long.  You've pushed Gigabytes of data with very few errors to show for it.  The 68/69 seconds of User Unavailable is odd, but that's the only thing I see wrong.

Something that is a bit odd is the 11.3 Gbytes down is paired with 8.23 GBytes upstream.  

Have you recently started doing online backups, streaming web or security cameras to the Internet or something similar that would cause your upload to be so high, compared to your download?  Why that can be a concern is that the Gateway doesn't do QoS within the Internet traffic it handles, so if you're sending 7Mbps of data upstream, then while that is going on, the traffic coming downstream isn't getting acknowledged in a timely fashion and things will start to break down.

@gr8sho -- there was no immediate resolution.  The service stabilized over night.  I went to bed (tossing arms in the air after getting disconnected from Cust Service), when I got up this morning, everything was back to normal.  No packet loss or diagnostic failures reported by the edge deivce.  But to your point, I think we can rule out the on-prem hardware.

Thanks for the analysis, @JefferMC .  I've not started using any online backups, so I'm not sure I can account for the differential.  We use VPN service when connecting to corporate networks, if that encapsulation impacts the counter imbalances.  But for the most part, our usage profile is heavily on the download side.  

One reason the problem might not show up in the counters is that if it is a port issue, and we can't even get a hostname resolved, those sessions never get established, so data downloads (i.e. for streaming) never materialize.  So established sessions seem fine, but resolving and establishing new sessions fails.  Make sense?  I originally thought this was a DNS issue.

The issue you describe sounds just like what I've been experiencing with a few places in AR.  When the issues are occurring I can use some of the built in tests on the arrisbgw210 and it will show lost packets, but the service will run ok for some time after the forced reboots ATT does.  I was finally able to get a tech to sit on the phone with me for about 10 minutes after he rebooted the modem and got him to also see packet loss when running tests from the modem.  He scheduled a dispatch, but it isn't the first dispatch for this particular case.  Another one with the same behaviors has had 3 or 4 dispatches with the modem being replaced on the last one.  Can't confirm that finally fixed it as that office has been closed since the modem replacement.  

Sorry I'm not contributing to your resolution, but I wanted to let you know someone else has been feeling the same struggle. 

Most people's usage is heavily weighted downstream.  For it to be so balanced, especially with a typical asymmetric bandwidth account, is very odd.  If you are participating in online meetings with your webcam on, that might help explain it.  If you're constantly in VPNs, then the following probably won't work.  I wanted to suggest you open a CMD window and do a "tracert 8.8.8.8".  In the output, the first hop should come back 192.168.1.254,  The second hop will be your Gateway router at AT&T.  Do a PING -t to that IP address and just leave it running.  The latency should typically be around 20-25 ms.  Hide the window if you want.  When you're having any sort of issue, pop up this window and see what this looks like.  Are you getting errors?  What's the latency?

@JefferMC spot-on.  The latency to that IP (just beyond the local gw) seems to settle at about 23ms.  There are currently no errors, but I will keep this shell script for use when the issue re-emerges.