Certain ipv6 hosts hang on certain machines on the network

EDIT: Fixed by switching internet service provider ¯\\(°_o)/¯ I have this annoying issue where some of the machines on my network will hang when looking up connecting to certain IPv6 addresses, in particular the ones from Google (but I've also seen issues with others when on the IPv6-enabled network). sadbox $ ping -6c1 -W10 fonts.googleapis.com PING fonts.googleapis.com(arn09s11-in-x0a.1e100.net (2a00:1450:400f:807::200a)) 56 data bytes --- fonts.googleapis.com ping statistics --- 1 packets transmitted, 0 received, 100% packet loss, time 0ms sadbox $ ssh happybox Last login: Wed Apr 25 10:28:48 2018 from 192.168.1.140 happybox $ ping -6c1 -W10 fonts.googleapis.com PING fonts.googleapis.com(arn09s11-in-x0a.1e100.net (2a00:1450:400f:807::200a)) 56 data bytes 64 bytes from arn09s11-in-x0a.1e100.net (2a00:1450:400f:807::200a): icmp_seq=1 ttl=55 time=25.6 ms --- fonts.googleapis.com ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 25.675/25.675/25.675/0.000 ms Both give the exact same answers for host fonts.googleapis.com and nslookup fonts.googleapis.com, and for the relevant parts of dig output. Some times the hosts will be connected to just fine on sadbox though, and some times I get a response if I wait long enough. It's not just ping that's affected, I noticed it because Firefox was waiting forever for fonts.googleapis.com on some site. The http://test-ipv6.com/ site gives full scores on sadbox, and I can ping and connect to other IPv6 addresses (e.g. sixxs.net) just fine. All my computers are running **Xubuntu 17.10**, and the router is running **LEDE Reboot (17.01.4, r3560-79f57e422d)**. What could the issue be? How do I debug something like this? ----- EDIT: Contents of sadbox:/etc/nsswitch.conf, comment-lines removed: passwd: compat group: compat shadow: compat hosts: files mdns4_minimal [NOTFOUND=return] dns mdns4 networks: files protocols: db files services: db files ethers: db files rpc: db files netgroup: nis and comparing to happybox: sadbox $ diff /etc/nsswitch.conf hosts: files mdns4_minimal [NOTFOUND=return] dns myhostname EDIT2: Pinging hangs on the IP too, e.g. ping -6 2a00:1450:400f:807::200a, so this is not DNS, but an IPv6 transport issue (thanks Patrick Mevzek!). I'll get traceroutes once I'm back on the network. ----- # Traceroutes: From the router itself it works fine: sadbox $ ssh ruter traceroute6 fonts.googleapis.com traceroute to fonts.googleapis.com (2a00:1450:400f:807::200a), 30 hops max, 16 byte packets 1 2a02:fe0:c900:1::1 (2a02:fe0:c900:1::1) 4.753 ms 5.894 ms 5.709 ms 2 2a02:fe0::a:47:a:13:4 (2a02:fe0::a:47:a:13:4) 5.944 ms 8.850 ms 14.108 ms 3 ae11-0.poh-pe1.stv.no.ip.tdc.net (2a02:228:41:100::127:0:9) 26.056 ms 15.009 ms 14.018 ms 4 stkm3nqp7.se.ip.tdc.net (2001:6c8:40::20) 22.371 ms 22.220 ms 21.974 ms 5 peer-as15169.stkm3nqp7.se.ip.tdc.net (2001:6c8:81:2000::9) 24.504 ms 25.047 ms 28.025 ms 6 2001:4860:0:1344::1 (2001:4860:0:1344::1) 26.118 ms 23.686 ms 24.075 ms 7 2001:4860:0:1::15cf (2001:4860:0:1::15cf) 22.793 ms 24.702 ms 23.859 ms 8 arn09s11-in-x0a.1e100.net (2a00:1450:400f:807::200a) 25.886 ms 35.837 ms 36.413 ms and the other working computer gives about the same: sadbox $ ssh happybox traceroute6 fonts.googleapis.com traceroute to ���2>V (2a00:1450:400f:807::200a) from 2a02:fe0:c910:5120:8011:3799:6dc0:2b58, 30 hops max, 24 byte packets 1 2a02:fe0:c910:5120::1 (2a02:fe0:c910:5120::1) 1,123 ms 1,247 ms 3,537 ms 2 2a02:fe0:c900:1::1 (2a02:fe0:c900:1::1) 6,958 ms 9,566 ms 6,955 ms 3 2a02:fe0:0:a:47:a:13:4 (2a02:fe0:0:a:47:a:13:4) 9,985 ms 6,303 ms 8,719 ms 4 ae11-0.poh-pe1.stv.no.ip.tdc.net (2a02:228:41:100:0:127:0:9) 6,811 ms 6,304 ms 6,741 ms 5 stkm3nqp7.se.ip.tdc.net (2001:6c8:40::20) 25,882 ms 26,625 ms 27,079 ms 6 peer-as15169.stkm3nqp7.se.ip.tdc.net (2001:6c8:81:2000::9) 27,169 ms 28,319 ms 28,931 ms 7 2001:4860:0:1343::1 (2001:4860:0:1343::1) 46,643 ms 28,799 ms 25,976 ms 8 2001:4860:0:1::b23 (2001:4860:0:1::b23) 29,373 ms 37,344 ms 28,207 ms 9 arn09s11-in-x0a.1e100.net (2a00:1450:400f:807::200a) 27,72 ms 26,274 ms 27,764 ms But from sadbox not so good: sadbox $ traceroute6 fonts.googleapis.com traceroute to ���V (2a00:1450:400f:807::200a) from 2a02:fe0:c910:5120:95b:58:ab20:d31e, 30 hops max, 24 byte packets 1 2a02:fe0:c910:5120::1 (2a02:fe0:c910:5120::1) 4,095 ms 4,632 ms 1,558 ms 2 2a02:fe0:c900:1::1 (2a02:fe0:c900:1::1) 11,822 ms 15,528 ms 10,568 ms 3 2a02:fe0:0:a:47:a:13:2 (2a02:fe0:0:a:47:a:13:2) 8,823 ms 6,119 ms 10,068 ms 4 * * * 5 * * * 6 * * * 7 * * * 8 * * * 9 * * * 10 * * * 11 * * * 12 * * * 13 * * * 14 * * * 15 * * * 16 * * * 17 * – it just keeps giving me stars (also, what's with that ���V?). A sudo traceroute -6 -T from sadbox gives the same. The IPv4 traceroute -4 works fine from all machines, including sadbox. Here's mtr from a working machine: sadbox $ ssh happybox mtr --report fonts.googleapis.com Start: Tue May 1 20:41:57 2018 HOST: happybox Loss% Snt Last Avg Best Wrst StDev 1.|-- 2a02:fe0:c910:5120::1 0.0% 10 0.7 1.0 0.7 2.3 0.5 2.|-- 2a02:fe0:c900:1::1 0.0% 10 7.4 9.0 6.2 14.2 2.6 3.|-- 2a02:fe0:0:a:47:a:13:4 0.0% 10 7.4 8.0 5.6 11.9 1.9 4.|-- ae11-0.poh-pe1.stv.no.ip. 0.0% 10 7.8 8.2 6.2 10.9 1.3 5.|-- stkm3nqp7.se.ip.tdc.net 0.0% 10 29.3 26.8 24.8 29.4 1.6 6.|-- peer-as15169.stkm3nqp7.se 0.0% 10 25.9 40.6 24.6 135.7 34.8 7.|-- 2001:4860:0:1344::1 0.0% 10 63.9 35.4 26.1 63.9 13.8 8.|-- 2001:4860:0:1::15cf 0.0% 10 26.1 28.1 24.7 36.4 3.2 9.|-- arn09s11-in-x0a.1e100.net 0.0% 10 26.7 26.8 24.8 31.4 1.9 while poor sadbox still gets no profit: sadbox $ mtr --report fonts.googleapis.com Start: Tue May 1 20:35:59 2018 HOST: sadbox Loss% Snt Last Avg Best Wrst StDev 1.|-- 2a02:fe0:c910:5120::1 0.0% 10 1.9 27.3 0.6 134.6 55.7 2.|-- 2a02:fe0:c900:1::1 0.0% 10 96.6 33.6 6.0 96.6 32.1 3.|-- 2a02:fe0:0:a:47:a:13:2 0.0% 10 7.8 8.0 6.4 14.0 2.1 4.|-- ??? 100.0 10 0.0 0.0 0.0 0.0 0.0