On Sunday night (Monday March 28), there appeared to be a maintenance window in my area (Mesa AZ). The service went out and there was a recording when calling Cox indicating planned maintenance. The next morning, I decided to try enabling IPv6. It worked! I disabled my HE.net tunnel and reconfigured everything to use the Cox native IPv6 space and the delegated prefix on my LAN.
Approximately one week later (about 12:30AM, Monday April 4), connections started to randomly fail. It's almost as if there's a multi link LAG in Cox's network, and one of those links is blackholing traffic. I say this because it isn't ordinary packet loss. Connections fail completely (including retransmits) or the succeed without loss, which points to a source/dest addr/port hashing algorithm usually used on LAGs.
This loss only occurs on the delegated prefix. If I make connections to or from my router machine directly, everything works. Pings and traceroutes to and from various IPv6 nodes on the Internet either fail or succeed depending on source/dest.
Here's a traceroute from the router (everything works from here)
-[~:#]- traceroute6 google.com
traceroute to google.com (2607:f8b0:4007:808::200e), 30 hops max, 80 byte packets
1 2600:8800:ff06::1 (2600:8800:ff06::1) 10.196 ms 15.877 ms 16.673 ms
2 2001:578:803:107::100 (2001:578:803:107::100) 16.692 ms 17.855 ms 17.866 ms
3 2001:578:805:104::48 (2001:578:805:104::48) 16.648 ms 21.696 ms 20.693 ms
4 2001:578:1:0:172:17:249:33 (2001:578:1:0:172:17:249:33) 33.737 ms 32.770 ms 32.712 ms
5 2001:4860:1:1:0:58f5:: (2001:4860:1:1:0:58f5::) 33.921 ms 33.688 ms 33.679 ms
6 2001:4860::1:0:6b02 (2001:4860::1:0:6b02) 33.896 ms 26.734 ms 27.342 ms
7 2001:4860:0:1::4f5 (2001:4860:0:1::4f5) 21.844 ms 26.557 ms 26.510 ms
8 lax17s05-in-x0e.1e100.net (2607:f8b0:4007:808::200e) 22.907 ms 27.106 ms 27.939 ms
Here's the same traceroute from a LAN host with the delegated prefix (connections randomly don't work)
-[~:$]- traceroute6 google.com
traceroute to google.com (2607:f8b0:4007:808::200e), 30 hops max, 80 byte packets
1 2600:8800:2000:4231::1 (2600:8800:2000:4231::1) 0.288 ms 0.282 ms 0.277 ms
2 2600:8800:ff06::1 (2600:8800:ff06::1) 16.558 ms 16.558 ms 16.555 ms
3 2001:578:803:107:: (2001:578:803:107::) 14.575 ms 15.611 ms 15.611 ms
4 2001:578:802:104::14 (2001:578:802:104::14) 14.558 ms 13.645 ms 14.553 ms
5 * * *
6 2001:4860:1:1:0:58f5:: (2001:4860:1:1:0:58f5::) 26.030 ms 22.866 ms 21.927 ms
7 2001:4860::1:0:6b02 (2001:4860::1:0:6b02) 28.130 ms 22.158 ms 22.138 ms
8 2001:4860:0:1::4f5 (2001:4860:0:1::4f5) 26.712 ms 28.257 ms 28.264 ms
9 * * *
10 * * *
11 * * *
12 * * *
13 * * *
14 * * *
15 * * *
16 * * *
17 * * *
18 * * *
19 * * *
20 * * *
21 * * *
22 * * *
23 * * *
24 * * *
25 * * *
26 * * *
27 * * *
28 * * *
29 * * *
30 * * *
What are my options in raising this issue with Cox? I've tried phone support ("It should work, we're not blocking anything") and email (got a canned reply about Cox's IPv6 plans).
↧