Anyone seeing issues between Cogent & Level3 in NYC?
I have Sprint & Cogent for bandwidth. Everything has been humming along for a couple years just fine. Yesterday around 8:00AM my BGP session with Cogent flapped. Now, when my Cogent BGP is up I get 100% packet loss in level3 land. When Cogent BGP is down (i.e. I’m running solely on Sprint) Everything is fine.
I have an open ticket with Cogent. They say they have a ‘capacity issue’ with level3 that has been escalated to executive levels.
With Sprint & Cogent BGP UP
I see traceroutes showing traffic leaving me on Sprint but returning on Cogent (and failing at level3). I’m guessing it is the level3/cogent border
With Sprint UP & Cogent Down
I see trace routes showing traffic on to/from on Sprint just fine.
Anyone else having issues?
-Matt
Unfortunately Cogent has a lot of peering issues. We use them in our network blend and we have been having lots of problems with traffic outbound to Comcast. It looks like from South Bend, Indiana on Cogent to Chicago / Level 3 we are getting a very tiny amount of packet loss and a higher than 'normal' latency of 35ms+.
Where are you connected to Cogent at? And what destination are you going to on Level 3?
Best Regards,
Honestly from the Internet Health Report, I've noticed connections between Level3 and Cogent are red quite a bit.
http://www.internethealthreport.com/
Bad samples or peering issues could be the cause either way, but it's been ongoing for awhile.
Sincerely,
Eric Tykwinski
TrueNet, Inc.
P: 610-429-8300
F: 610-429-3222
Unfortunately Cogent has a lot of peering issues. We use them in our network blend and we have been having lots of problems with traffic outbound to Comcast. It looks like from South Bend, Indiana on Cogent to Chicago / Level 3 we are getting a very tiny amount of packet loss and a higher than 'normal' latency of 35ms+.
Yeah, I know they are always my secondary, never my primary
Where are you connected to Cogent at? And what destination are you going to on Level 3?
Boston (300 Bent) but I think they haul it to 1 Summer St
A bunch of sites fail but www.cnn.com is one that comes to mind.
Possibly related to their mass outage last night around 5:12am CST
(ticket number HD0000005596458). We're connected at their 427 S La
Salle POP in Chicago.
brandon
Cogent found the problem today. It took them 4 days to do a ‘show conf’ and see that an outbound access-list was applied to my interface by mistake during a ‘normal maintenance window at 8AM EST on Friday’
4 days of jumping through hoops to prove that the problem wasn’t on my network. grumble.
-Matt