Jul 22, 2023
4
0
10
I live near Austin and starting about a week ago I started getting intermittent packet loss (every 10-20 seconds) completely out of the blue. Ran multiple ping tests, and traceroutes, and have a ping plotter log showing that the packet loss doesn't start until it hops to an AT&T IP address whose IP lookup approximates to be in or near Austin. I also have video clips of me playing video games and packet loss in games coinciding with timed-out ping requests to different IPs as well as packet loss to the aforementioned AT&T IP address.

I'll attach the ping plotter log for reference but I am mainly making this post because I'm wondering if, despite the evidence otherwise, it truly is a problem on my end and I'm just crazy. I tried calling AT&T earlier and the customer support basically made me restart my modem and ran some speed tests on their end and then concluded the issue was fixed (spoiler alert: it's not) which is about how I expected the call to go. Much to my surprise (not), they said any packet loss would be an issue with my 3rd party router.

The slightly bright side of the situation is that for the most part, my connection is fine. Web browsing and streaming shows are mostly unaffected. However, as someone who plays a lot of video games, it is extremely noticeable, especially in FPS games. I have a novice understanding of networking so I would love to hear some more knowledgeable opinions on my ping plotter log or troubleshooting tips I could try. Thanks.

Ping Plotter Log
 
Solution
Your new traces don't show a lot new. You can't read a lot into a large spike in just one node. Just like the one in hop 5 show 100% loss. Most routers are configured to prefer actual traffic first than responding to ping packets. Some also have a limit on how many packets they respond to in a certain period of time...some are set to never respond. This is all to prevent denial of service attacks using ping commands since it is very easy to spoof the source address preventing tracking of a attacker.

What is key is in the trace is you are seeing very consistent 3-4% loss starting in hop 4 and then going all the way to the end with at least that much loss in every hop past it. That pretty much means the problem is in...
Try to post the ping plot results a different way it does not resolve for me.

The most common place to get loss is going to be on the connection between your house and att. This though is much more rare on fiber than on say a copper coax cable type of connection. You would normally see the problems start in hop 2. In your cases since you have a extra router it would likely be hop 3. This connection and the att modem/router itself are what the level 1 techs are trained to troubleshoot.

If your problem is past the "wire" between your house and the ISP it gets much more complex. In many cases the level 1 tech does not have access to the equipment. It also is made more complex because the traffic internal to att...ie when they test from the att office..may not follow the same data path as your traffic to servers. If you could test ping plot to some att address it would somewhat help. The other thing that makes it confusing is traceroute , pingplot, etc only look at the path from you TO the end device. The return path can actually follow a different set of routers. You would have to be able to run traces in both directions and be able to test from intermediate routers to see if you were losing the packet going from your machine to the test server or response coming back from the server to you. There are some testing tools that will attempt to measure this but it doesn't really help with getting it fixed since only someone with access to the intermediate routers can really see anything useful.

I am not sure how you proceed with this. Years ago when I had a similar issue I had a friend who now worked for the ISP and could use his internal contacts and could confirm to them that I actually knew what I was talking about. Turned out to be a bad interface board in one of their main routers that only had issues with certain blocks of IP addresses.

When the problem is in the main infrastructure of a ISP and it is not major enough for their monitoring tools to detect it these problems are hard to get fixed.

This is why you really hope it is on the connection coming to your house. These are much simpler to get fixed.
 
Jul 22, 2023
4
0
10
Try to post the ping plot results a different way it does not resolve for me.

The most common place to get loss is going to be on the connection between your house and att. This though is much more rare on fiber than on say a copper coax cable type of connection. You would normally see the problems start in hop 2. In your cases since you have a extra router it would likely be hop 3. This connection and the att modem/router itself are what the level 1 techs are trained to troubleshoot.

If your problem is past the "wire" between your house and the ISP it gets much more complex. In many cases the level 1 tech does not have access to the equipment. It also is made more complex because the traffic internal to att...ie when they test from the att office..may not follow the same data path as your traffic to servers. If you could test ping plot to some att address it would somewhat help. The other thing that makes it confusing is traceroute , pingplot, etc only look at the path from you TO the end device. The return path can actually follow a different set of routers. You would have to be able to run traces in both directions and be able to test from intermediate routers to see if you were losing the packet going from your machine to the test server or response coming back from the server to you. There are some testing tools that will attempt to measure this but it doesn't really help with getting it fixed since only someone with access to the intermediate routers can really see anything useful.

I am not sure how you proceed with this. Years ago when I had a similar issue I had a friend who now worked for the ISP and could use his internal contacts and could confirm to them that I actually knew what I was talking about. Turned out to be a bad interface board in one of their main routers that only had issues with certain blocks of IP addresses.

When the problem is in the main infrastructure of a ISP and it is not major enough for their monitoring tools to detect it these problems are hard to get fixed.

This is why you really hope it is on the connection coming to your house. These are much simpler to get fixed.
Thank you for your detailed response. I think I actually saw some responses you had made on super old threads I was looking at.

I have attached the photo that you weren't able to access. Unfortunately, it seems as though the worst-case scenario you described may be true. Packet loss doesn't start until the 4th hop which appears to be an intermediary ATT IP. I will run some tests on the first and last ATT addresses in the path that are outside the line to my house and post them here soon.

Weirdly enough, a few hours after I had called them yesterday I saw an ATT van driving around my neighborhood looking at all kinds of different stuff. I'd like to believe it was more than just a coincidence because lately, people other than myself have also been complaining on neighborhood forums so maybe we ruffled some feathers.

BhGoZnLvdfo.png
 
Jul 22, 2023
4
0
10
Okay so here I have attached two short ping tests, one to the first address outside of my internet line (71.149.77.196) and the other just to att.com. I tried to run tests to other intermediary ATT IPs but it would simply just have 100% packet loss when it got to the destination (maybe they block direct ping requests?)

The plot to att.com looks very similar to the Google one I previously provided. Packet loss starts at hop 4 and continues throughout with seemingly no issues within my network. However, when I ping the aforementioned address from my first paragraph (which is where the packet loss starts) packet loss starts to show up at almost exact intervals to my ATT Modem.

Just in case that is confusing, the 192.168.1.254 address is that of the black router box ATT has provided and it receives connection directly from the fiber line. That provides a WAN connection to my Unifi USG (192.168.2.1) which then connects to various switches and access points in my house.

I ran both these tests at the same time so I don't know if that would've affected anything. If this gives any insight into a possible problem I could fix on my end please let me know.


CHCNYDozsr9.png
YRSKy1LQW9b.png
 

Ralston18

Titan
Moderator
Also bear in mind that the current heat waves and severe storms may be causing problems.

Problems for both power and communications lines.

Especially if the various tests indicate problems outside of your network and perhaps AT&T as well.

People staying inside and online (video gaming) may be affecting bandwidth availability.

Just my thoughts on the matter.
 
Your new traces don't show a lot new. You can't read a lot into a large spike in just one node. Just like the one in hop 5 show 100% loss. Most routers are configured to prefer actual traffic first than responding to ping packets. Some also have a limit on how many packets they respond to in a certain period of time...some are set to never respond. This is all to prevent denial of service attacks using ping commands since it is very easy to spoof the source address preventing tracking of a attacker.

What is key is in the trace is you are seeing very consistent 3-4% loss starting in hop 4 and then going all the way to the end with at least that much loss in every hop past it. That pretty much means the problem is in hop 4.

If you get lucky you will get a tech that does more than follow their script and understands this problem and can escalate it to a higher level person.

This pretty clearly shows there is some issue in the att network on that node....but even with all the fancy cisco certs I have I don't really know how you talk your way past the robot level 1 techs.
 
Solution
Jul 22, 2023
4
0
10
Also bear in mind that the current heat waves and severe storms may be causing problems.

Problems for both power and communications lines.

Especially if the various tests indicate problems outside of your network and perhaps AT&T as well.

People staying inside and online (video gaming) may be affecting bandwidth availability.

Just my thoughts on the matter.
Thank you for this insight - not really things that I considered. It did start around the time that the "heat wave" we are currently going through started.

This pretty clearly shows there is some issue in the att network on that node....but even with all the fancy cisco certs I have I don't really know how you talk your way past the robot level 1 techs.
I definitely know what you're talking about regarding the robot techs haha. I'll probably see if it comes to pass but if it persists for another week or two I'll start trying my luck with bothering ATT. I'll go ahead and flag bill001g's latest message as the answer as there's nothing else to really talk about.

Thank you both for your time and insight, it was very helpful. Take care.