Sudden spike in discos

[hawkeye61]

Usually, I'll see about 3-6 discos per month.

I'm now up to 6 in two days. Lost UDP switched to TCP.

As suggested, here's my pingplot data after 10 mins. I'm not quite sure what I'm looking at, other than XX% packet loss sounds really bad.



Can anyone tell me what's going on here?

[The Fugitive]

Skuzzy might have to run a ping back to you. When you have loses in the next to the last hop like that he's says its a sign there might be trouble in the return trip. You could send Skuzzy a note, and he WILL answer with in 24 hours. If not, he didn't get it or the reply got lost.

[Skuzzy]

I see the problem on the return trip.  Level3 is in your path, for the return.  That pretty much insures high packet loss, particularly during primetime as Lvel3 is Netflix's primary carrier.  It is simply a very oversold ISP.

[hawkeye61]

Ok, so where does that leave us/me?

[ToeTag]

I called level 3's NOC one time pretending to be a direct customer the guy found me out but still was helpful.  They told me that ping plotter was not accurate and to use the windows version of doing a trace route.  It still read as packet loss and they basically said we will look into it.  After that My signal was still low and slower than most but the packet loss and disco's seemed to decrease.

[Skuzzy]

Register a complaint with you ISP that Level3 is dropping packets at the gateway into your ISP's network and push them for a resolution.

It will be difficult, as Level3 continues to oversell thier network and as they are a cheap connection to the Internet, many cable ISP's use them.

I have registered complaints with AT&T about this before and will continue to do so.  Level3 has redefined what "over-selling" and Internet connection is.

[hawkeye61]

Three discos in 30 minutes tonight. First after just logging on. Second after about 20 mins of flight. Third after logging on again. Awesome.

Will call cable company to see if there are larger issues at work here, but I'm still stuck on why this was not a problem until recently.

[The Fugitive]

Three discos in 30 minutes tonight. First after just logging on. Second after about 20 mins of flight. Third after logging on again. Awesome.

Will call cable company to see if there are larger issues at work here, but I'm still stuck on why this was not a problem until recently.

A storm in the area could have messed up a connection point, the cable company could have just switched over to another company for the relaying of their data connections any number of things could have changed outside of your little corner of the world to effect your connection. It took a couple of weeks for the connection for those in England to settle down after THE wedding of the Prince and Princess.

[james]

My pingplot looks just like this one as well. It's not just your little corner of the world. Just called level3 and they said it was att. Ill try to get ahold of someone over there tomorrow but really what can they do?

[Skuzzy]

If you are anywhere near the east coast of the U.S., you will see more sporadic problems due to all the damage done to the telecommunications infrastructure by the recent weather.

Anytime a major portion of any country is hit with large scale telecommunications damage it tends to impact a much larger percentage of people outside of the confines where the actual damage took place.

[hawkeye61]

Skuzzy,

Now up to disco 13 since beginning of tour 140.

Talked to my ISP. Had a guy come out to look at the hardware side - he tweaked a couple things, but nothing major - still having the problem. I've sent some of the pingplot data to my ISP and they're having a look at things to see if they can identify any problems on their end - using my pingplot data, as well as your recommendation of looking at Level3's gateway on the return trip.

Up at the top, my original post, it looks like the vast majority of the packet loss is occurring at hops:

9
11 - ggr1.dlstx.ip.att.net
13 - mdf001c7613r0003-gig-10-1.dal1.attens.net

I understand you said second to last hop may indicate problems in the return trip. What about hops 9 and 11? Is that still the outgoing data being lost? If so, that is very likely outside my ISP's influence. Does HTC use AT&T? Is this what you were referring to before when you said you logged complaints with them? If so, what have the responded to you?

Any chance you could post your pingplot data back to my machine so that I could see it too? Might be helpful if I could include that in the info I'm sending my ISP.

Thanks.

[Skuzzy]

All your problems are on the return path.  I cannot get a reliable trace done as it seems to be changing routes every minute ot two.

[hawkeye61]

All your problems are on the return path.  I cannot get a reliable trace done as it seems to be changing routes every minute or two.

What would be causing that?

Understand ALL of my problems are on the return path? It doesn't seem to look that way. Not trying to debate with you, but to better understand what's going on. The more I know, the better I can present my issues to my ISP.

ISP technician supposed to come back out today and have another look. Here's hoping.

[Skuzzy]

The problem with any type of traceroute/Ping Plot is it can only show you the data path TO to the server, from your computer location.  When an error occurs on the return path, which is different than the data path TO the server, these utilites show the error on the last router in the chain.

On occasion the last few hops of a trace to us will show the ethernet switches, when AT&T, occasionally enables ICMP ECHO reponse on them.  Switches cannot route, so the last router in the chain ends up showing any errors on the asynchronous return path.  It is simply the nature of how traces work and the limitations therein.

[hawkeye61]

18! 18 Discos! Ah! Ah! Ah! (Plus a few that didn't get counted because it discoed as soon as I entered the MA.)

Seriously, wtf is going on here? I know I'm not the only one having these problems either. Several squad mates are suffering from chronic discos too.