IPV6 stops to work some (variable) time after boot

[satollo]

Hi, the machine is a Centos (centos-release-7-4.1708.el7.centos.x86_64) in Hetzner datacenter. The problem is the IPV6 stack, see below a curl to google.com:

Code: Select all

[root@www ~]# curl -vvvv www.google.com
* About to connect() to www.google.com port 80 (#0)
*   Trying 2a00:1450:4001:81f::2004...
* Connection timed out
*   Trying 216.58.206.4...

BUT if I reboot the server for a while the IPV6 connection works perfectly. To me it sounds like a network problem in the datacenter but the support, of course, says "NO".
Could be a misconfiguration of the centos image? It contains only nginx, mariadb and php-fpm, postfix and monit, all in the latest version. If something works for a while and then stops, I could image the problem is raised by some late executed command, like by a cron (but actually the cron has nothing special but a mariadb dump for backup).

Is there some debug command I can launch to better check where the problem is?

Thank you very much, Stefano.

[avij]

traceroute6 might show you something.