[ntp:questions] Problem starting ntpd

Paul Croome Paul.Croome at softwareag.com
Wed Aug 20 16:15:10 UTC 2003


I'm having trouble starting ntpd on a Sun SPARCstation IPC with
Solaris 7. The
/etc/inet/ntp.conf file looks like this:

 server mode 130 # Meinberg C51
 fudge time1 0.026
 fudge time2 0.001
 peer timenet2
 driftfile /etc/ntp.drift
 logfile /var/adm/ntp.log

Starting the daemon with 'ntpd -dddd' mostly, but not always, gives:

interface <lo0> OK
interface <le0> OK
bind() fd 4, family 2, port 123, addr, flags=1
flags for fd 4: 0202
bind() fd 5, family 2, port 123, addr, flags=0
flags for fd 5: 0202
bind() fd 6, family 2, port 123, addr, flags=1
flags for fd 6: 0202
create_sockets: ninterfaces=3
interface 0:  fd=4,  bfd=0,  name=wildcard,  flags=0x1
              sin=  bcast=,  mask=
interface 1:  fd=5,  bfd=0,  name=lo0,  flags=0x4
              sin=  mask=
interface 2:  fd=6,  bfd=0,  name=le0,  flags=0x1
              sin=  bcast=, 
init_io: maxactivefd 6
getnetnum given, got (7f7f0800)
peer_clear: at 0 assoc ID 0
input_handler: if=2 fd=6 length 48 from 9dbda017
newpeer:> mode 3 vers 4 poll 6 6 flags 21 1 ttl
130 key 00000000
getnetnum given, got (7f7f0800)
getnetnum given, got (7f7f0800)
resolving timenet2
config: timenet2 1 4 6 10 0 0 0 *
getnetnum given, got (a1497ff)
peer_clear: at 0 assoc ID 0
newpeer:> mode 5 vers 4 poll 6 10 flags 1 4
ttl 0 key 00000000
authtrust: keyid 0000ffff life 1
report_event: system event 'event_restart' (0x01) status 'sync_alarm,
sync_unspec, 1 event, event_unspec' (0xc010)
Segmentation Fault (core dumped)


The last line in /var/adm/ntp.log is:

20 Aug 15:52:53 ntpd[12475]: parent died before we finished, exiting


If I delete the first 3 lines of the ntp.conf file, ntpd starts up and
fine. However, I'm trying to run a high-stratum NTP server with a
radio clock...

Also, if I try repeatedly to start ntpd, it ultimately does kick in,
after 100..150 attempts :-(

My interpretation of this is that the IPC hardware, being somewhat
slow, is not
able to initialize and start the refclock driver within the time that
the parent
process considers to be reasonable. I've looked for a timeout in the
source code
(4.1.1b) but without success.

Any ideas? [Buying new hardware is not an option!]


Paul Croome
Paul.Croome at softwareag.com

More information about the questions mailing list