unable to create socket on eth0 (210) for ::ffff:192.168.1.4#123

[Thomas Schweikle]

Binary package hint: ntp

ntpd[738]: unable to create socket on eth0 (210) for ::ffff:192.168.1.4#123
ds ntpd[738]: failed to initialize interface for address ::ffff:192.168.1.4
ntpd[738]: bind() fd 22, family AF_INET6, port 123, scope 2, addr ::ffff:192.168.1.4, mcast=0 flags=0x11 fails: Invalid argument

Hmmmm?

ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: ntp 1:4.2.4p8+dfsg-1ubuntu6
ProcVersionSignature: Ubuntu 2.6.35-24.42-generic-pae 2.6.35.8
Uname: Linux 2.6.35-24-generic-pae i686
Architecture: i386
Date: Wed Jan 5 11:33:13 2011
InstallationMedia: Ubuntu-Server 10.10 "Maverick Meerkat" - Release i386 (20101007)
NtpStatus:
 remote refid st t when poll reach delay offset jitter
 ==============================================================================
  192.168.1.255 .BCST. 16 u - 64 0 0.000 0.000 0.001
ProcEnviron:
 LANG=de_DE.UTF-8
 SHELL=/bin/bash
SourcePackage: ntp
mtime.conffile..etc.ntp.conf: 2011-01-04T17:58:09.340110

[James Page (james-page)]

Thank you for taking the time to report this bug and helping to make Ubuntu better.

This looks like a potential bug/configuration problem with the IP setup of your Ubuntu installation; ::ffff:192.168.1.4 is a ipv4 address that has been mapped to the ipv6 address space.

Please could your provide more information about your networking setup; specifically any ipv6 related configuration you may have.

Thanks

[Thomas Schweikle (tps)]

Networking setup is quite simple: the machine in question does only have one network interface:

auto eth0
iface eth0 inet static
  address 192.168.1.4
  netmask 255.255.255.0
  gateway 192.168.1.1

iface eth0 inet6 static
  address ::ffff:192.168.1.4
  netmask 96
  gateway ::ffff:192.168.1.1

ntp is bound to all interfaces, all ipv4, ipv6

[James Page (james-page)]

I have been able to confirm this bug using the information supplied by the Thomas in the last update.

Although ntp generates error messages every 5 minutes, it does respond to queries on the ipv6 address.

This may be related to bug 700492 (although the error message is slightly different they both relate to ipv6 network handling in ntp).

[Thomas Schweikle (tps)]

Yes, both bugs are related. ntp assumes 0::ip-v4-address, ff:ff::ip-v4-address and ip-v4-address the same (they are not, but this is what ntpd thinks of). Since ip-v4-address is already opened, opening 0::ip-v4-address and ff:ff::ipv4-address will fail, because ntpd just strips off the higher bits and uses ip-v4-address instead.

[Thomas Schweikle (tps)]

Aaarrgggg: has to read ::ffff:ip-v4-address and ::ip-v4-address, not ff:ff::ip-v4-address and 0::ip-v4-address!

[Thomas Schweikle (tps)]

Fixed in a later release of ntpd