Northeastern US User Routed to Europe for Free Plan?
Starting around last night, I have noticed laggy loading of pages on my sites CDNed with Cloudflare my ISP (RCN, AS6079) in Boston. A curl to the /cdn-cgi/trace shows colo=MAD which I interpret to be that it is routed to Madrid. Latter I saw the colo field changes to LHR and AMS occasionally. This morning, I have my friend in the Boston metro (another ISP, Crown Castle, AS46887) test the same for me (with my site and another friend's), the trace also shows LHR as the colo field. I also tested with my server in Digital Ocean NYC2 (with IPv6 and IPv4) and LHR is the colo.
From my experience, traffic from my ISP is routed to BOS or EWR so the cross-Atlantic routing makes little sense to me. I understand that there is no guarantee for the "optimal routing" but I am still curious if this is intended or is misconfiguration or lingering effect from the route leaking incident yesterday afternoon?
Another thing that puzzles me is that MTR to the IP address our domains resolve to shows a perfectly normal path (ISP -> Boston IX -> Cloudflare). Also paid plan (e.g., ChatGPT) and 1.1.1.1 are not laggy and is showing colo=BOS for the trace.
Example curl:
ā ~ curl https://dzh.wtf/cdn-cgi/trace
fl=990f76
h=dzh.wtf
ip=146.115.88.xxx
ts=1755799379.000
visit_scheme=https
uag=curl/8.7.1
colo=LHR
sliver=none
http=http/2
loc=US
tls=TLSv1.3
sni=plaintext
warp=off
gateway=off
rbi=off
kex=X25519
8 Replies
What do you see with https://cloudflare.com/cdn-cgi/trace
fl=1024f33
h=cloudflare.com
ip=146.115.88.254
ts=1755799550.256
visit_scheme=https
uag=Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:141.0) Gecko/20100101 Firefox/141.0
colo=IAD
sliver=none
http=http/2
loc=US
tls=TLSv1.3
sni=plaintext
warp=off
gateway=off
rbi=off
kex=X25519MLKEM768
Yeah, it seems to be only Free plan (which unfortunately my friends and I are on)
yeah that's been happening the past couple days
it's not related to any of the other things
it's not exactly wrong. this is a thing that can happen sometimes. it's happening more than usual this week
Oh. Thanks! just want to give a heads up in case nobody is aware.
What makes it confusing is the MTR results shows "correct" path with low latency (but HTTP conversation definitely shows lag) so I initially thought it is just server "thought" it is in MAD/LHR.
Cloudflare has a ton of cool tools for load balancing, like Plurimog https://blog.cloudflare.com/meet-traffic-manager/
Plurimog does the same thing, but instead of talking between machines, Plurimog talks in between data centers and points of presence. If a request goes into Philadelphia and Philadelphia is unable to take the request, Plurimog will forward to another data center that can take the request, like Ashburn, where the request is decrypted and processed. Because Plurimog operates at layer 4, it can send individual TCP or UDP requests to other places which allows it to be very fine-grainedAn alternative to move just enough traffic rather then dropping all free traffic for a PoP
much of my week has involved what's going on here
Relieved to know BOS->MAD weirdness is ongoing and not something I messed up š
- sending my best to you folks
I call it MADness to my friends