We're implementing a new ipv6 router, would anybody be interested in beta testing it?
mercutio: I'm intrigued. What's involved?
(just my /48 gets moved/announced from the new router, and I update my end?)
brycec: We'd just port your vlan over.
You wouldn't change your end.
Cool. Should I expect much interruption/stability? (I mean, yeah it's a beta test, but *how* beta is it? :p)
Either way, I have a /64 (viable for beta-beta) and /48 (ideally late-beta) I'm happy to test with.
brycec: a little potentially
that's why looking for beta testers first. :)
it shouldn't be greatly unstable.  it's like running a beta OS :)
I'm pretty confident it's solid
Sounds good. My /64 does very, very little traffic while my /48 disruptions would be noticed but wouldn't be show-stopping.
it's not like running on beta hardware, it's more that there is a chance for issues because less proven stable rather than predictably unstable.
That's all the endorsement I need :)
What's next, need me to email? PM?
We'll be terminating Cogent there as soon as they get their heads out of their asses and actually set up our new session
("terminating" is really terrible language choice when crossing from telecom back to the real world "Oh you're killing the contract?")
there's also killing a session etc, yeah, it can be odd ;)
brycec: nothing, mercutio will move your VLAN
then there's "server daemons"
That's what she said!!
Haha yeah
your /64 and /48 are on the same vlan?
BryceBot: no
Oh, okay... I'm sorry. 'then there's "server daemons"'
mercutio: no, completely different accounts in fact.
2607:f2f8:abf0::/64 and 2607:f2f8:a650::/48
brycec: ah, so you want to move just your /64 now, or the /48 as well?
mercutio: Yeah both accounts, more traffic is better right?
sure :)
(both now, I wasn't clear)
ping me if any issues
(Assuming I can connect to Freenode :P *rimshot*)
heh
you can always connect over ipv4 if ipv6 issues :)
ok moving your /64 shortly
brycec: do you have a test IP that you'd like me to check works?
vps3.cobryce.com
(That's on the /48)
ah on the /64
(Had to look that one up) 2607:f2f8:abf0::253 should work
cool, i'll move your vlan now brycec
and check that that works :)
it's moved but i can't reach from international
hangon :)
100% packet loss?
Yeah, unreachable it seems. Coming from HE.
yeh it's from he.net i'm seeing it not work too
can you ping the gateway from there?
because the gateway can ping you
mercutio: Yes, that :253 host is able to ping 2607:f2f8:abf0::1. My home connection is NOT able to ping 2607:f2f8:abf0::1
it's odd, i can see the traffic going to you just no return
like it's lost it's default router
but still can see the gateway
can you see the default gw still looks sane?
weird i just saw some dns request come through
Its default route is definitely set and working, and it was able to ping an IPv6 address of Google.
oh
But when I try to reach that host from my home HE tunnel, nada.
yeh hangon
from ARP FRA doesn't work either, in case that helps you
ok better now brycec ?
mercutio: Seems to work
sorry about that
Beta(TM)
ironically it was the old router being the issue
it didn't load the route into fib from rib
that is the kernel routing table didn't match the userlevel routing table
still doesn't work from FRA
oh fIorz ?
[...]
11  r2.lax.arpnetworks.com (2607:f2f8:0:7::165)  179.839 ms  185.234 ms  175.512 ms
12  * * *
brycec: are you allowing icmp?
or maybe I am pinging the wrong address?
2607:f2f8:abf0::253 is the one that I don't get any responses from, ::1 and :a650::3 do respond
only abf0 has moved, and ::1 is gateway, that is why i was asking bryce :)
i can see traffic going through to it, but not coming back
I'm sorry, I stepped away earlier once everything was working. Yes actually, the device blocks all traffic by default including ICMP packets.
cool
brycec: how is your latency? :)
I haven't noticed any changes.
heh
mercutio: Did you port the /48 I gave you too? I didn't notice any disruptions (I expected Freenode would've disconnected)
brycec: not yet :)
it'll be moved tonight most likely
Cool
That one I use much, much more so I'll notice any changes in latency there.
cool.
(The /64 you moved earlier is used basically-never, rarely anything "interactive" it's just there to be dualstack to a few sites that are 99.99% IPv4-only)
Does your normal route go via any2ix?
I have no idea, and `mtr` can't seem to resolve any of the hops to DNS names so I still don't know. (looking at the un-moved network/host)
(Even traceroute6 fails to resolve them, blah)
that is so common with ipv6
I'm blaming Debian this time. The versions of mtr and traceroute on my home desktop have no issues with PTR for most of the hops
2001:504:13::211:15
2001:504:13::211: would be the any2ix
15 is facebook :)
Ah yes I see that /36 (32?) in my mtr
 2. AS???   2001:504:13::1a
cool
oh oops i maent 13
(mtr couldn't even determine its AS number)
i stripped off the last ip heh
2001:504:13::/48
so yeah any2ix should be migrated onto it soon too
Just checked and both the moved and unmoved networks, going to google.com, are going over ANY2IX.
yeh google will, ISP traffic and other further away traffic is less likely
people like google, facebook etc peer a lot :)
well ISP traffic in US I should say is less likely to
In that case, testing from my home HE.net tunnel to both the "moved" and "not yet moved" networks goes via HE.net -> ge-100-0-0-13.r00.lsanca07.us.ce.gin.ntt.net
(And off the top of my head, I only have HE.net tunnels to test from)
he.net goes via any2ix
it'll be reverse path going over ntt
ntt is terminated on a different router than any2ix
And whadya know, running mtr the opposite way (towards my home connection) does show ANY2IX on both old and new v6 routers
How's it looking now?