AT&T Fiber: Slow Internet (Latency / Packet Loss) started on or right after 6/30

@adac-solutions We must live somewhere nearby, or at least be affected by same part of connection. I woke up early on 8/10 (around 5 am) and although I don't have auto-monitoring, I test it regularly and that was the first time the test went completely smoothly.

I will continue monitoring and if you could also - I definitely feel that during the day network is not nearly as stable.

Also, if you can add few details about your monitoring setup? I have Raspberry Pi laying around and I'd love to set it up for that as well.

Thanks!

@adir11 I have a private Github repo I've been managing my monitoring infrastructure out of that I plan to release at some point in the near future.  Right now it's a bit of a mess and I haven't kept it clean from personal details because it was private.  So I'll have to go through a pass or two before I make it public.

I'm super swamped today at work, but I'll try to get some rough ideas and a network layout diagram up here when I can.

In other news, I kept my appointment with AT&T yesterday.  The tech that came to the house said he stopped by and talked to the techs that were repairing the cut lines (apparently comcast hit them doing their own work) and said they replaced a card for our subdivision.  The timing of when the card was replaced and a possible update I feel like doesn't line up with 2:30AM EDT, but perhaps software updates or the like to this "card" he refers to did.  Grasping at straws for closure here.

I've thought about throwing my Graphite-loading PowerShell scripts into GitHub, but I don't know that anyone would be that interested.  I've run most of them on Windows and on Linux (Debian, but not Raspbian).

Assuming he's not just blowing smoke, the card would be the OLT card at the CO.

@JefferMC thanks for the information on the OLT card.  He didn't say the card by name, but he did mention the CO.  Yeah I don't know if he was blowing smoke or not.  At the end of the day it's fixed and as I had been saying all along it wasn't my house or my equipment.

I plan to continue to keep my monitoring running moving forward as a stead state when things go bad.

There's definitely value in providing that information.  I'm actually not running Raspbian on my RasberryPi 4Bs.  I'm running ubuntu server 22.04 edition to keep it consistent with what I'm running on my linux servers at the house.  Make the configuration simpler. Though netplan was a bit new to me over the way I'm using to setting up multiple nics and routes.

Other than a couple of brief play sessions with Red Hat, I've spent most of my Linux time on a Pi, providing help/support for a local FIRST FRC team who used them as robot coprocessors.  I loaded Debian on my folding server I use for various things (like Graphite) because it would keep me from dealing with differences; basically I went for compatibility in the opposite direction. :-)  But what I know about networking on Linux would still probably fit in a thimble.

I have scripts for scraping counts off the various BGW interfaces and reporting them to STATSD (ported from earlier scripts that do the same thing for MRTG), do pings against various servers, etc.

One thing that @ATTHelp mentioned that was interesting is that having your AT&T gateway plugged into a power strip can cause power fluctuations that manifest as connectivity issues.

Sorry, I’m going to rant here:

This makes me want to scream. As an electrical engineer, and a set top designer for a decade, and dozens of years in electronics, I hear this all the time from AT&T. Nothing in my training would explain why being plugged into a power strip, basically a parallel group of sockets, (or a UPS for that matter) would generate any new “power fluctuations” that can effect connectivity differently than being plugged into the wall. Power strips are passive devices with maybe a one time surge protector device, but that’s all.

If the gateway power supply is so pathetic that having something plugged in “along side” the device can cause problems, then you should also be telling people not to plug anything into any other outlet on the same circuit breaker, or maybe even the same house, because all those outlets are basically a large power strip in your home anyway, a group of parallel connected sockets. 

I have run dozens of tests with multiple graphing power line monitors running at the same time on different outlets in a strip and around the house and have NEVER seen power fluctuations on a power strip outlet that differ at all from fluctuations on any single wall outlet in the same house, allowing for a bit of noise amplitude reduction based on distance from the source of the noise, but mostly negligible. Devices on a power strip showed zero difference from other outlets on the same circuit breaker branch in the same room. And I added large load switching on and off the strip to generate fluctuations. The other nearby wall outlets were identical. 

In addition, your power company has plenty of power fluctuations all the time in their delivered power. Houses on the same transformer will see fluctuations from each other as well for things like AC starting up, washers starting, electric stoves and ovens switching on, electric dryers starting, pool equipment, hair dryers, etc, etc.  And a good UPS could help protect against some of those fluctuations because they usually contain some power conditioning circuits to smooth out line noise. 

I really wish AT&T would stop saying this. Seems like just another reason to blame the customer for something when they don’t know what the real problem is. 

AT&T, if you have an engineering explanation for this power strip statement, I would live to read it. 

end of rant. 

I've always wondered if maybe this had originated back in the day when people would plug their DSL line into the RJ11 ports on some power strips.  I can imagine that some of those devices weren't kind to marginal lines, thus, "power strips are bad, mkay".

I've stopped doing it lately, but whenever ATTHelp posted this crap, I liked to respond with a link to the Official AT&T video on how to install the BGW320 gateway.  It specifically says that the manufacturer recommends that you plug it into a power strip. 

Obviously nobody that controls the boilerplates pays any attention.  Same with the one that tells people to make sure their version of Adobe Flash should be up to date.