0

Secondary Anycast IPv6 timing out

Morning,

The secondary anycast IPv6 address (2a07:a8c1::) has been timing out for me for some time causing issues with dns resolution.

anexia-lon (IPv6) is returning "error" on ping.nextdns.io, which I presume is causing the timeout issue.

Diag here: https://nextdns.io/diag/a5785d00-e441-11ee-94b4-fb6155f40355 (I'm aware it shows that I'm unconfigured, that's intentional until this is resolved.)

Cheers,

-Will

11 replies

null
    • Will_Tisdale
    • 8 mths ago
    • Reported - view

    I contacted my ISP about this and they raised a ticket with Anexia yesterday evening and the routing issue is now resolved.

    Although anycast still isn't great because the latency on 2a07:a8c1:: is spiking quite drastically during the daytime so it's clearly overloaded. 

    2a07:a8c0:: has packet loss and latency spikes within the NTT network.

    Can you look into this please?

    • Will_Tisdale
    • 8 mths ago
    • Reported - view

    anexia-lon no longer has spikes of 200ms latency and has been mostly at about 12ms for the past 24 hours - I presume someone fixed whatever was wrong with it, thanks.

    vultr-lon still has intermittent high latency and intermittent packet loss - originating from 2001:728:0:5000::bea which is in the NTT network.

      • NextDNs
      • 8 mths ago
      • Reported - view

      can you please send a https://nextdns.io/diag when the issue on lon happens?

      • Will_Tisdale
      • 8 mths ago
      • Reported - view

       I would, but as of about 40 mins ago (according to my monitoring graph) I'm being routed to zepto-ams instead of vultr-lon now.

      Doing a mtr on the vultr-lon IPv6 address as of now shows a completely different route to yesterday (no NTT) and no packet loss. And a very nice 9ms ping too...I presume you've been busy changing stuff.

      • Christophe_Yayon
      • 8 mths ago
      • Reported - view

       me too.

      No more NTT.

      $ traceroute -I 45.90.30.0
traceroute to 45.90.30.0 (45.90.30.0), 30 hops max, 60 byte packets
 1  router1.nbux.org (192.168.2.7)  0.108 ms  0.134 ms  0.117 ms
 2  80.10.238.153 (80.10.238.153)  1.503 ms  1.587 ms  1.611 ms
 3  lag-10.necls17z.rbci.orange.net (193.249.213.173)  12.560 ms  12.547 ms  12.534 ms
 4  * ae110-0.ncann201.rbci.orange.net (193.253.84.242)  12.880 ms  12.929 ms
 5  ae45-0.nrlyo201.rbci.orange.net (193.252.101.10)  12.884 ms  12.872 ms  12.935 ms
 6  * * *
 7  * * *
 8  * * *
 9  * * *
10  * * dns2.nextdns.io (45.90.30.0)  12.767 ms

$ traceroute -I 45.90.28.0
traceroute to 45.90.28.0 (45.90.28.0), 30 hops max, 60 byte packets
 1  router1.nbux.org (192.168.2.7)  0.118 ms  0.119 ms  0.179 ms
 2  * 80.10.238.153 (80.10.238.153)  1.454 ms *
 3  lag-10.necls17z.rbci.orange.net (193.249.213.173)  12.666 ms  12.647 ms  12.653 ms
 4  ae110-0.ncann201.rbci.orange.net (193.253.84.242)  12.797 ms  12.777 ms  12.800 ms
 5  ae45-0.nrlyo201.rbci.orange.net (193.252.101.10)  18.055 ms  18.040 ms  18.054 ms
 6  ae42-0.nridf101.rbci.orange.net (193.252.101.214)  12.744 ms * *
 7  * ae41-0.noidf001.rbci.orange.net (193.252.98.101)  12.766 ms  12.552 ms
 8  * * *
 9  acorus.th2-1.hopus.net (37.77.34.83)  20.050 ms * *
10  * * *
11  * * *
12  * * *
13  edge01.pa2.par.netbarista.com (193.168.204.0)  13.419 ms  13.404 ms *
14  dns1.nextdns.io (45.90.28.0)  14.021 ms  13.187 ms  15.104 ms
      • Will_Tisdale
      • 8 mths ago
      • Reported - view

       to follow up, 2a07:a8c0:: is being routed to vultr-lon now instead of zepto-ams. The vultr-lon ping times look far better than it was when it was going via NTT,  packet loss has dropped to nothing too. I think you've fixed the issues, thanks.

      nas.z3n.uk (2001:8b0:6401:82a0:0:ffff:a0a:a05) -> 2a07:a8c0:: (2a07:a8c0::)                                2024-03-20T21:49:50+0000
Keys:  Help   Display mode   Restart statistics   Order of fields   quit
                                                                                           Packets               Pings
 Host                                                                                    Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. router.z3n.uk                                                                         0.0%    61    2.4   2.7   2.1   9.1   1.2
 2. g.gormless.thn.aa.net.uk                                                              0.0%    61   10.2  10.1   8.8  13.6   1.0
 3. ITAF.bar2.Brussels1.Level3.net                                                        0.0%    61    9.9  10.0   8.8  14.8   1.1
 4. ge-5-3-6-181.bar2.Brussels1.Level3.net                                                0.0%    61   10.0  10.8   9.4  14.7   1.0
 5. lo-0-v6.ear1.London1.Level3.net                                                      13.1%    61    9.5  11.0   9.4  16.7   1.4
 6. CHOOPA-LLC.ear1.London1.Level3.net                                                    0.0%    60   10.5  15.8   9.6  56.3   9.9
 7. ethernetae2-sr2.lon3.constant.com                                                     0.0%    60   10.8  13.1  10.0  45.7   5.6
 8. ethernetswp25-ds1-u3-r51b4-b.lon3.constant.com                                        0.0%    60   10.0  10.9   9.6  17.7   1.3
 9. (waiting for reply)
10. 2a07:a8c0::                                                                           0.0%    60    9.5  10.7   9.2  17.9   1.5
      • Will_Tisdale
      • 8 mths ago
      • Reported - view

       I'm back being routed via NTT for vultr-lon and the latency spikes are back.

      The latency is originating in the NTT network, as it was the other day. And it seems to be the same IP address where it starts - 2001:728:0:5000::bea

      There's also some occasional packet loss, but nowhere near as bad as it was the last time I was being routed via NTT.

      Managed to capture a mtr output whilst a latency spike was present - pretty obvious that it's originating in NTT:

      nas.z3n.uk (2001:8b0:6401:82a0:0:ffff:a0a:a05) -> 2a07:a8c0:: (2a07:a8c0::)                                2024-03-23T09:18:26+0000
Keys:  Help   Display mode   Restart statistics   Order of fields   quit
                                                                                           Packets               Pings
 Host                                                                                    Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 2001:8b0:1111:1111:0:ffff:51bb:b98e                                                   0.0%   107    2.2   2.4   2.0   8.3   0.6
 2. 2001:8b0:0:53::133                                                                    0.0%   107   11.3  11.2  10.3  12.7   0.4
 3. 2001:728:0:5000::682                                                                  0.0%   107   11.6  11.4  10.0  20.1   1.0
 4. 2001:728:0:5000::681                                                                  0.0%   107   10.4  12.5  10.2  23.8   2.9
 5. 2001:728:0:5000::bea                                                                  0.0%   107   63.2  19.0  10.8  67.5  14.2
 6. 2001:19f0:fc00::a45:346                                                               0.0%   107   37.4  17.4  11.1  69.8  11.9
 7. 2001:19f0:fc00::a45:aa                                                                0.9%   107   39.9  15.2  10.6  68.0  10.9
 8. (waiting for reply)
 9. 2a07:a8c0::                                                                           0.0%   106   58.9  14.3  10.4  68.9  11.9

      Running a diag at the moment, which will take forever as it always hangs for ages on the IPv6 anycast traceroutes whilst it's timing out.

      Will post when it is complete although given that the spikes are intermittent I don't think it will show you anything useful.

      Also attached monitoring comparing vultr-lon with anexia-lon.

      • Will_Tisdale
      • 8 mths ago
      • Reported - view

      https://nextdns.io/diag/ef9ee0b0-e901-11ee-a098-c74f46017d77

      • NextDNS_Network
      • 8 mths ago
      • Reported - view

       Hi, we've enabled zepto-lon's LINX / LONAP peering for AANet.

      The traceroute you provided shows that latency issues only occur after entering the Vultr network, and we've checked that both Vultr and Anexia use LINX for AAnet, so it could be that Vultr's LINX port is congested and causing problems.

      Please let us know if the problem persists.

      • Will_Tisdale
      • 8 mths ago
      • Reported - view

       thanks for that. Congestion makes sense, it's either that or an unhappy switch. The constant latency spikes on 2a07:a8c0:: have stopped after the move to zepto-lon as the graph below shows. I'll keep an eye on it but it seems to be fine now.

    • Christophe_Yayon
    • 8 mths ago
    • Reported - view

    Same issue for me (from France). I already opened a bug report in this forum...

    It seems to be better since yesterday 17h35 (GMT+1). Did you change something ?

Login to reply

Content aside

  • 8 mths agoLast active
  • 11Replies
  • 122Views
  • 4 Following