On Tue, Mar 07, 2017 at 09:08:07PM +0000, Unman wrote: > On Tue, Mar 07, 2017 at 09:56:23PM +0100, 'Antoine' via qubes-users wrote: > > On Mon, Mar 06, 2017 at 04:31:31PM -0800, Andrew David Wong wrote: > > > >> Filed a bug report: > > > >> > > > >> https://github.com/QubesOS/qubes-issues/issues/2674 > > I have the same problem with Fedora 23, Debian 8 and Debian 9: > > > > = Fedora 23 = > > [user@work ~]$ grep PRETTY /etc/os-release > > PRETTY_NAME="Fedora 23 (Workstation Edition)" > > [user@work ~]$ cat /etc/resolv.conf > > nameserver 10.137.2.1 > > nameserver 10.137.2.254 > > [user@work ~]$ dig +short gov.uk @10.137.2.1 > > 23.235.33.144 > > 23.235.37.144 > > [user@work ~]$ dig +short gov.uk @10.137.2.254 > > ;; connection timed out; no servers could be reached > > > > = Debian 8 = > > user@cloud:~$ grep PRETTY /etc/os-release > > PRETTY_NAME="Debian GNU/Linux 8 (jessie)" > > user@cloud:~$ cat /etc/resolv.conf > > nameserver 10.137.2.1 > > nameserver 10.137.2.254 > > user@cloud:~$ dig +short gov.uk @10.137.2.1 > > 23.235.33.144 > > 23.235.37.144 > > user@cloud:~$ dig +short gov.uk @10.137.2.254 > > ;; connection timed out; no servers could be reached > > > > = Debian 9 = > > user@Email:~$ grep PRETTY /etc/os-release > > PRETTY_NAME="Debian GNU/Linux 9 (stretch)" > > user@Email:~$ cat /etc/resolv.conf > > nameserver 10.137.2.1 > > nameserver 10.137.2.254 > > user@Email:~$ dig +short gov.uk @10.137.2.1 > > 23.235.33.144 > > 23.235.37.144 > > user@Email:~$ dig +short gov.uk @10.137.2.254 > > ;; connection timed out; no servers could be reached > > > > Do you have an advise how to remove 10.137.2.254 from the list of > > default name servers? > > Probaly more relevant would be for you to discover why the first > nameserver isnt reachable or isnt responding. > With multiple entries they are queried in the order given, so if the > first is working correctly the second entry wont be hit. > > Thats the real problem.
I have understood why I have this problem. On my LAN, the DNS recursive server (unbound) has a blacklist: it refuses to answer queries for tracking/ad domains. The problem is that when a program receives a "REFUSED" packet from its DNS query, it tries to solve the same host on the second DNS server in resolv.conf. I can see the pattern clearly using tcpdump: Query -> fast answer REFUSED -> Query on the second DNS server -> no answer. On the DNS resolver: # grep facebook unbound-blacklist.conf local-zone: "facebook.com" refuse on any Qubes VM: $ host facebook.com 10.137.2.1 Using domain server: Name: 10.137.2.1 Address: 10.137.2.1#53 Aliases: Host facebook.com not found: 5(REFUSED) $ host facebook.com 10.137.2.254 [... 10s ...] ;; connection timed out; no servers could be reached $ host facebook.com Host facebook.com not found: 5(REFUSED) $ ping facebook.com [... 10s ...] ping: facebook.com: Temporary failure in name resolution I do not understand why this second DNS server is populated in all Qubes VM. Is there a simple way to configure only 1 DNS server? Antoine -- You received this message because you are subscribed to the Google Groups "qubes-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to qubes-users+unsubscr...@googlegroups.com. To post to this group, send email to qubes-users@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/qubes-users/20170308225516.3coxi2iautyfbfuj%40fedora-23-dvm. For more options, visit https://groups.google.com/d/optout.
signature.asc
Description: PGP signature