Re: High latency for IPv6 on netbsd-8

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Mon, 30 Jul 2018 21:38:01 +0700
    From:        Gua Chung Lim <gua.chunglim%gmail.com@localhost>
    Message-ID:  <20180730143801.GA82%gmail.com@localhost>


  | > tcpdump -s 2000 -w DUMP-FILE -i wm0
  | It gave me a binary log. But how to read it?

tdpdump -s 2000 -r DUMP-FILE   (that was in the earlier message).

  | > ping6 -n 2001:3c8:9007:1::21 (munnari.oz.au - one of its addresses)
  | I shall ping6 between 3:00-4:00pm GMT.

Sorry, it is now 16:xx UTC and I only just saw this e-mail - that was
just after 16:00, since then I have been building this reply (Iwas away
from my keyboard for a couple of hours earlier.)

  | But note that ping6 while it is unable, it maybe only a short term.
  | I cannot estimate, sometimes 5 minutes or sometimes 30 minutes.

That will be OK.   Since it is clearly not an RA issue, the next most
likely thing with a period like that would be something routing related
(or possibly some dynamic firewall somewhere.)

  | After that period ping6 becomes enable without any notification.
  | So I'm not sure whether at the time you check, it is enable or not.

I will do a tcpdump just like above.   Then you tell me what time you tried
and it failed, and I'll check what was happening at that time.

After that you casn just not attempt any more pings (to here).

But I'd appreciate a smaller window than an hour if possible, that would
make a VERY big dump file if I just capture everything, and I would prefer
not to filter too much (except I can limit it to v6) in case that causes a 
false negative (if my filter caused my tcpdump not to save your packets...)

  | > I could also try a ping6 from here to your host if I know more or less
  | > exactly what time to do it (when your host has just rebooted and things
  | > are not working as they should).
  | Do you mean pinging my router or my host? My host sits behind NAT.

NAT for IPv6 ?   Why?    There's no point checking IPv4, that's working, right?

If you are really attempting IPv6 NAT for some weird reason, stop...

  | Or shall I forward the port from router to my host?

ICMP has no ports.   So that's not possible.

  | (I remember when I was using 6bone, anyone can ping6 me behind LAN without port forwarding.)
  | Anyway my host is 2405:9800:b550:2939:f234:69d6:e0bf:8ebf/64
  | or 2405:9800:b550:2939:8638:35ff:fe48:5720/128 and
  | my router is 2405:9800:b550:2939:8ee1:17ff:fe1f:d1c7.

Right now I cannot get to you at all (what is the router's V6 address on the
interface facing the internet) ?

IPv6 packets look to go from here to the edge of the core network (or somewhere
near that) and then no further, which would indicate that there is no available
route to 2405:9800::    Of course, thatmight be a local problem near my end, so
I will try again later.

  | https://pastebin.com/6K8s2QWi
  | https://pastebin.com/AKFbrmYk
  | https://pastebin.com/kH7mc1ZB
  | https://pastebin.com/uvbbQDsq

I see nothing very revealing there - just that when it is failing the IPMPv6
echo packets being sent is going up (it seems, with just one output it is
hard to be sure) but no replies are coming back (and that when it is
working you are getting replies).    That all looks like  the host is doing
what it should, and most likely the problem is elsewhere (at the router,
or further into the internet).

If you can use some other host on the same LAN, and send ICMPv6 pings
from it, which work, at the same time (or very close) to when pings from
the NetBSD-8 host are failing, that would be useful info.   If that works,
the IPv6 address of that host would be useful to know as well.

  | > You can add `nodhcp6` to dhcpcd.conf to disable DHCP6 entirely.
  | I have done it.
  | % tail -n 2 /etc/dhcpcd.conf
  | logfile /var/log/dhcpcd.log
  | nodhcp6
  | Now ping6 is usable even if DHCP6 is disabled entirely.

Are you saying that with that config it works, or that it does not work ?

  | No, it was using dhcpdc and rtsol on -7.
  | I haven't used dhclient for long, probably -5.

Oh sorry, obviously I misinterpreted when you said you would try
"going back to dhclient" (which I do not think would help anything.)

kre