this is not a bug report, just some information about an issue Aidan Gauland reported on the Debian bugtracker.
He was using a firejail sandbox with --net=eth0 and was losing internet connectivity inside the sandbox (but not outside).
We figured out hostname resolution suddenly stopped working. He is using rdnssd on the host, which is a tool that autodiscovers IPv6 nameservers. This was overwriting the resolv.conf with a nameserver detected on the local network.
The content looked like this: nameserver fe80::c6e9:84ff:1234:abcd%eth0
But as there was no eth0 interface inside the sandbox (instead there is a eth0-nnnnn interface), the nameserver was no longer reachable.
He also suggests to document this behavior on the known issues page.
Hi,
this is not a bug report, just some information about an issue Aidan Gauland reported on the Debian bugtracker. He was using a firejail sandbox with --net=eth0 and was losing internet connectivity inside the sandbox (but not outside). We figured out hostname resolution suddenly stopped working. He is using rdnssd on the host, which is a tool that autodiscovers IPv6 nameservers. This was overwriting the resolv.conf with a nameserver detected on the local network. The content looked like this:
nameserver fe80::c6e9:84ff:1234:abcd%eth0
But as there was no eth0 interface inside the sandbox (instead there is a eth0-nnnnn interface), the nameserver was no longer reachable.He also suggests to document this behavior on the known issues page.