|
ATT datacenter router problem - again ping plot |
Hey folks, Anyone else in Austin area seeing massive packet loss upstream? Looks like they have 2 IP addresses that are dying. |
|
|
|
Anon3c7a9
Anon
2024-Jul-9 1:41 am
I see the exact same thing (in Round Rock). I thought it was something internal to me but saw on MTR and other tools there was quite a bit of loss upstream. Not all sites are impacted. It's hit or miss. I'd open a ticket, but I'm not sure they'd do anything. |
|
scooby Premium Member join:2001-05-01 Schaumburg, IL
1 recommendation |
to robsquattro
Network gear makes responding to icmp (ping) packets the lowest priority if it has other tasks so those two hops in the middle are not anything to worry about. It looks like you have very minor packet loss to www.google.com (about 1.7%). |
|
|
Looks like it was fixed sometime last night. Those routers are now at 0% packet loss |
|
|
Anon3c7a9
Anon
2024-Jul-9 9:51 am
Much better today. Mine wasn't 1.7%, I had loss as high as 20+ percent to some sites. Some of the bigger collaboration/video conferencing sites as well, which made meetings terrible. It appears much much better today. |
|
|
|
Well I spoke too soon. They are back to problematic. |
|
|
to robsquattro
Would a simple continuous ping command for 30sec+ repeat these issues or does it need to run a lot longer? |
|
|
Oh, interesting. The route my packets are taking just updated. The 2 bad IP addresses are no longer in the route and I'm back to normal. I use ping plotter which does a better job at showing where and when packet loss occurs along the path. |
|
robsquattro |
Looks like we're in for another bad night. Latency is spiking again. |
|
tito79 join:2010-03-14 Port Saint Lucie, FL
2 recommendations |
tito79
Member
2024-Jul-9 8:04 pm
Probably the hurricane that passed |
|
Astyanax Premium Member join:2002-11-14 Melbourne, FL
1 recommendation |
to robsquattro
said by robsquattro:Looks like we're in for another bad night. Latency is spiking again. It's the tides from the storm surge from the hurricane rolling in and out of the data center. It sucks a redundant router or switch out to sea once in a while. |
|
|
to tito79
said by tito79:Probably the hurricane that passed Probably not. |
|
tito79 join:2010-03-14 Port Saint Lucie, FL |
tito79
Member
2024-Jul-10 2:04 pm
Did they fix it ? |
|
|
Not yet. Still higher than normal latency and a couple routers with terrible packet loss |
|
2 recommendations |
said by robsquattro: a couple routers with terrible packet loss Get used to this. Some routers will not respond to every timed-out ICMP request. |
|
THX join:1999-10-16 Saint Louis, MO
2 recommendations |
THX
Member
2024-Jul-11 11:35 am
said by JefferMC:Get used to this. Some routers will not respond to every timed-out ICMP request. 100% this. ICMP can be wildly unpredictable. You will sometimes see results like this... |
|
1 recommendation |
Two different things. A reporting loop is not the same as a failure to return a TTL error.
BTW, 2a01:110::/31 is owned by Microsoft. |
|
1 recommendation |
to Astyanax
Its just cleaning your packets |
|