Anyone else seeing appalling performance with HE's tunnels the past 12 hours?
(Figured others here use HE too, might have seen same.)
(Or at least heard about it from a list they follow)
yes
Specifically I'm seeing over 50% packet loss to my gif's endpoint (tserv1.sea1) itself, and that really doesn't help the tunneled traffic.
painfully high latency and loss
Well always nice when I'm not alone :)
must be all the netflix ipv6 angst ;)
lol
(Not in my household - Netflix pulled that bullshit and I null-routed their IPv6 prefix)
brycec: I am curious how you figured out all the prefixes they use
Google :P
dont they use akamai and/or amazon to host their content
They self-host most of their stuff actually
in addition to ISP/Partner Netflix connect
you used pfsense at your network edge I presume to null-route?
http://www.dslreports.com/forum/r30792233-ipv6-Netflix-now-also-blocking-HE-net-IPv6-Tunnels
a huge chunk of their v6 stuff is in AWS, though
which sucks
so i did the dns proxy
which is annoying, because it breaks DNSSEC
i wish bind rpz could filter that crap
(Currently down to 8% loss and falling, yippee)
0 loss in the last 200 seconds, and normal ping times, this feels much better
Oh bother, packet loss once again.
brycec: i haven't been having packet loss issues with he.net tunnel to arp
well smokeping doesn't show such
it may be bad to other places
...why would anyone with native v6 tunnel to ARP?
it's tracing to arp
from he.net
so lax to lax
i'm also monitoring the germany arp location too though
which is also fine from lax
Ohhh I misunderstood, got it now.
cool
Well it's not HE's network that seems to be the issue but rather an overloaded (DDoS'd, etc) POP, tserv1.sea1.he.net
ahh ok
i only recently changed from san jose to lax termination
Was getting loss+latency with plain old v4 traffic to that host.
but they actually have a few locations, maybe sjc would work better for you?
actually i'm not sure what location it was
now that i think about it
because their main dc is fremont
yeah it was fremont
much of a muchness i suppose
mercutio: I'm pretty sure Seattle is my closer (network, and geographic) POP https://dl.dropboxusercontent.com/u/3167967/screenshot_2016-07-07_11-38-50.png
doesn't help if network is lousy to there ;)
traceroute *to* the POP was fine too
oh i thought you said it was bad with ipv4 too
mercutio: Right, at hop 9
tserv3.fmt2.ipv6.he.net is where i was
"ping -4 tserv1.sea1.he.net" so to speak
heh wy route to tserv1.sea.he.net is via he.net in lax
Really? Goes via Level3 for me ARP -> tserv1.sea1.he.net
from new zealand i mean
ah
so i can't really test sea well