MacOS – Mac OS X Mountain Lion – DNS resolving uses wrong order on VPN via dial-up connection

dnsmacosvpn

I am using a MacBook with Mac OS X 10.8.2 and connect to my company's network via VPN. Everything works great when establishing the VPN connection via LAN or WLAN. However, when I use a dial-up connection (Huawei HSDPA USB Stick) host names are not correctly resolved in applications (e.g. Web-Browser). Command line tools like host name will correctly resolve the IP address, ping name will not resolve.

Using scutil --dns I dumped the DNS configuration when connecting via WLAN vs. dial-up. There is a notable difference in the lookup order:

connecting using WLAN:

resolver #1
  nameserver[0] : 192.168.80.10
  nameserver[1] : 192.168.80.24
  if_index : 6 (ppp0)
  reach    : Reachable,Transient Connection
  order    : 100000

resolver #2
  nameserver[0] : 192.168.80.10
  nameserver[1] : 192.168.80.24
  if_index : 6 (ppp0)
  reach    : Reachable,Transient Connection
  order    : 200000

resolver #3
  domain   : local
  options  : mdns
  timeout  : 5
  order    : 300000

resolver #4
  domain   : 254.169.in-addr.arpa
  options  : mdns
  timeout  : 5
  order    : 300200

resolver #5
  domain   : 8.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 300400

resolver #6
  domain   : 9.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 300600

resolver #7
  domain   : a.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 300800

resolver #8
  domain   : b.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 301000

DNS configuration (for scoped queries)

resolver #1
  nameserver[0] : 192.168.1.1
  if_index : 4 (en0)
  flags    : Scoped
  reach    : Reachable,Directly Reachable Address

resolver #2
  nameserver[0] : 192.168.80.10
  nameserver[1] : 192.168.80.24
  if_index : 6 (ppp0)
  flags    : Scoped
  reach    : Reachable,Transient Connection

The ppp0 connection is the VPN connection. As you can see, two servers are connected and they answer correctly on the command line and in applications.

Connecting via UMTS:

resolver #1
  nameserver[0] : 139.7.30.126
  nameserver[1] : 139.7.30.125
  if_index : 6 (ppp0)
  reach    : Reachable,Transient Connection
  order    : 100000

resolver #2
  nameserver[0] : 192.168.80.10
  nameserver[1] : 192.168.80.24
  if_index : 7 (ppp1)
  reach    : Reachable,Transient Connection
  order    : 100000

resolver #3
  nameserver[0] : 192.168.80.10
  nameserver[1] : 192.168.80.24
  if_index : 7 (ppp1)
  reach    : Reachable,Transient Connection
  order    : 200000

resolver #4
  domain   : local
  options  : mdns
  timeout  : 5
  order    : 300000

resolver #5
  domain   : 254.169.in-addr.arpa
  options  : mdns
  timeout  : 5
  order    : 300200

resolver #6
  domain   : 8.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 300400

resolver #7
  domain   : 9.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 300600

resolver #8
  domain   : a.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 300800

resolver #9
  domain   : b.e.f.ip6.arpa
  options  : mdns
  timeout  : 5
  order    : 301000

DNS configuration (for scoped queries)

resolver #1
  nameserver[0] : 192.168.80.10
  nameserver[1] : 192.168.80.24
  if_index : 7 (ppp1)
  flags    : Scoped
  reach    : Reachable,Transient Connection

resolver #2
  nameserver[0] : 139.7.30.126
  nameserver[1] : 139.7.30.125
  if_index : 6 (ppp0)
  flags    : Scoped
  reach    : Reachable,Transient Connection

This time, ppp1 is the VPN connection and ppp0 is the UMTS connection. From the response times of the commands (using the non-existing hostname foo.bar.local) I infer that ping uses the first resolver chain, where as host uses the scoped query configuration. ping takes 5 seconds to return "Unkown host", hostgets back immediately. I assume ping runs into the 5 second timeout of the mdns resolver.

In order to fix my problem with the broken DNS lookups when dialing in via VPN over modem I need to change the order of the resolvers. So far I have not found a way of doing this.

Any ideas welcome.

Best Answer

I had the same problem on my Mac, and after fixing it I have figured out that it was caused by FortiClient (VPN client). Even when FortiClient was disconnected - it's DNS still appeared in the scutil.

The solution for me was:

scutil
> list ".*DNS"

This will show you a list of all DNS configs, that will look something like:

subKey [0] = State:/Network/Global/DNS <br>
subKey [1] = State:/Network/MulticastDNS<br>
subKey [2] = State:/Network/OpenVPN/DNS<br>
subKey [3] = State:/Network/OpenVPN/OldDNS<br>
subKey [4] = State:/Network/PrivateDNS<br>
subKey [5] = State:/Network/Service/forticlientsslvpn/DNS <br>

To check each of them run: (until you find the problematic one)

> get key_name
> d.show

…and to fix it run:

> get key_name
> d.remove ServerAddresses
> set key_name

This is how it looked on my machine:

> get State:/Network/Service/forticlientsslvpn/DNS 
> d.show
<dictionary> {
  ServerAddresses : <array> {
    0 : 192.168.30.6
    1 : 192.168.30.15
  }
  SupplementalMatchDomains : <array> {
    0 :
  }
  SupplementalMatchOrders : <array> {
    0 : 100000
  }
}
> d.remove ServerAddresses
> d.show
<dictionary> {
  SupplementalMatchDomains : <array> {
    0 :
  }
  SupplementalMatchOrders : <array> {
    0 : 100000
  }
}
> set State:/Network/Service/forticlientsslvpn/DNS
> exit
Related Question