Ticket ID: SIXXS #556305 Ticket Status: Remote Problem PoP: nlams04 - Stipte B.V. (Amsterdam)
High packet loss in tunnel statistics
Shadow Hawkins on Tuesday, 31 July 2007 01:26:05
Hello,
I have static 6in4 tunnel to nlams04 PoP with following parameters:
Tunnel Name niva.net.ua-nslams04
PoP Name nlams04
PoP IPv4 213.204.193.2
Your Location Chernigov, Ukraine
Your IPv4 195.60.175.13
IPv6 Prefix 2001:960:2:591::1/64
PoP IPv6 2001:960:2:591::1
Your IPv6 2001:960:2:591::2
I use Debian GNU/Linux 3.1 (oldstable) on router if that matters in my case.
Problem: there is very high packet loss (up to 100% at any time of the day) reported on Loss Statistics Graph on my Tunnel Information page at SixXS. I'd like to know why it is so. ICMPv6 packets are not blocked by firewall at my router.
According to my little testing there are no problems with connectivity from my router to the PoP but there is so huge packet loss reported on site. And because of it there was few days logged as "tunnel is down" which is not so good for credits.
Here is some results of ping6 commands to understand my point better:
1. ping6 from the my router to the PoP's end of tunnel with small packets:
$ ping6 -c 200 -i 0.3 2001:960:2:591::1
PING 2001:960:2:591::1(2001:960:2:591::1) 56 data bytes
64 bytes from 2001:960:2:591::1: icmp_seq=1 ttl=64 time=62.7 ms
...
64 bytes from 2001:960:2:591::1: icmp_seq=200 ttl=64 time=57.6 ms
--- 2001:960:2:591::1 ping statistics ---
200 packets transmitted, 200 received, 0% packet loss, time 60495ms
rtt min/avg/max/mdev = 55.179/61.126/94.912/6.493 ms
2. ping6 from the my router to the PoP's end of tunnel with big packets:
$ ping6 -c 200 -i 0.3 -s 1500 2001:960:2:591::1
PING 2001:960:2:591::1(2001:960:2:591::1) 1500 data bytes
1508 bytes from 2001:960:2:591::1: icmp_seq=1 ttl=64 time=72.4 ms
...
1508 bytes from 2001:960:2:591::1: icmp_seq=200 ttl=64 time=72.6 ms
--- 2001:960:2:591::1 ping statistics ---
200 packets transmitted, 199 received, 0% packet loss, time 60488ms
rtt min/avg/max/mdev = 68.034/72.575/118.542/5.048 ms
3. ping6 from the one random IPv6 enabled host in Germany to the my end of tunnel:
$ ping6 -c 200 -i 0.3 2001:960:2:591::2
PING 2001:960:2:591::2(2001:960:2:591::2) 56 data bytes
64 bytes from 2001:960:2:591::2: icmp_seq=1 ttl=52 time=67.8 ms
...
64 bytes from 2001:960:2:591::2: icmp_seq=200 ttl=52 time=72.5 ms
--- 2001:960:2:591::2 ping statistics ---
200 packets transmitted, 199 received, 0% packet loss, time 60882ms
rtt min/avg/max/mdev = 66.396/71.597/97.882/5.817 ms
Here is the traceroute6 from that host to the my end of tunnel:
$ traceroute6 -n 2001:960:2:591::2
traceroute to 2001:960:2:591::2 (2001:960:2:591::2) from 2001:780:0:25::80, 30 hops max, 16 byte packets
1 2001:780:0:25::1 0.642 ms 0.382 ms 0.328 ms
2 2001:780:40:3::2 0.793 ms 0.828 ms 0.579 ms
3 2001:780:40:a::2 17.561 ms 3.987 ms 4.653 ms
4 2001:668:0:3::2000:221 4.077 ms 4.02 ms 4.197 ms
5 2001:668:0:2::1:442 4.335 ms 4.249 ms 6.177 ms
6 2001:668:0:2::4b0 4.645 ms 4.414 ms 4.453 ms
7 2001:668:0:2::1:402 7.225 ms 7.302 ms 8.3 ms
8 2001:668:0:2::1:122 11.176 ms 12.769 ms 11.492 ms
9 2001:668:0:2::630 12.047 ms 11.617 ms 12.165 ms
10 2001:7f8:1::a501:2634:2 11.938 ms 11.952 ms 11.728 ms
11 2001:960:226::134 11.966 ms 11.961 ms 12.101 ms
12 * * 2001:960:800::2 16.014 ms
13 2001:960:2:591::2 69.578 ms 67.785 ms 72.813 ms
4. Results from the ping6 tool on web site http://www.berkom.blazing.de/tools/ping.cgi to the my end of tunnel:
PING 2001:960:2:591::2(2001:960:2:591::2) 56 data bytes
64 bytes from 2001:960:2:591::2: icmp_seq=1 ttl=58 time=93.5 ms
64 bytes from 2001:960:2:591::2: icmp_seq=2 ttl=58 time=91.2 ms
64 bytes from 2001:960:2:591::2: icmp_seq=3 ttl=58 time=94.6 ms
64 bytes from 2001:960:2:591::2: icmp_seq=4 ttl=58 time=88.2 ms
64 bytes from 2001:960:2:591::2: icmp_seq=5 ttl=58 time=89.9 ms
--- 2001:960:2:591::2 ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4035ms
rtt min/avg/max/mdev = 88.231/91.550/94.693/2.347 ms
As I can see there is no packet loss in all cases at least at this time of day. But why the huge packet loss is logged on the site?
Maybe the problem is in big RTT time (usually > 55 ms)? But it's inevitable for me because of large network and geographical path from the PoP to me (the north of Ukraine).
--
With best regards,
Alexey Molchanov,
AVM106-RIPE
State change: remoteproblem
Jeroen Massar on Tuesday, 31 July 2007 08:40:38
The state of this ticket has been changed to remoteproblem
High packet loss in tunnel statistics
Jeroen Massar on Tuesday, 31 July 2007 08:40:57
Read the FAQ on why your tunnel doesn't ping.
Posting is only allowed when you are logged in. |