[ntp:questions] xntpd woes on non-global Solaris 10 zone

cedricmarvin at yahoo.com cedricmarvin at yahoo.com
Tue Dec 20 22:23:12 UTC 2005


  I am working with a Solaris 10 system with "zones" setup and I have
been tasked with getting xntpd running in a non-global (not zone 0)
zone which can't make ntp_adjtime system calls because its not zone 0.
  Why is this happening ?  Consolidation of legacy servers and the
primary admin of this Solaris 10 system does not want to put the legacy
dns server (which was on a standalone Sparc 5 box) into zone 0 as a
virtual network interface.
  The Zone 0 part of this Solaris 10 box is talking ntp to an external
server over it's network interface, so the system clock is okay.

  So, I just want the xntpd in the non-global zone to listen for
network connections and hand out whatever it has for the time (which is
the Zone 0 time which is ntp sync'd).

  The system is running xntpd 3-5.93e+sun which came with Solaris 10

  In ntp.conf I have:

   fudge stratum 0

  But even though it is using the local clock, /var/adm/messages keeps

   Dec 20 14:02:55 dns01 xntpd[5380]: [ID 373468 daemon.error]
     adj_frequency: ntp_adjtime failed: Not owner
   Dec 20 14:08:15 dns01 last message repeated 5 times

  I looked in the source for xntpd 3-5.93e and it seems that even if
you have the local clock as primary, it still wants to make the
ntp_adjtime system call.

  A couple of questions:
   - Any way to stop the ntp_adjtime via some config parameter ?
   - Do newer versions of xntpd also have this issue ?

  I'll put in a newer ntp daemon, but the in brave new world of Solaris
10 services I'm not totally confortable doing this (but I guess I have
to get bloody at some point).

  Thanks in advance for any advice and help !

  - Cedric

More information about the questions mailing list