ntp issues time not synchronized

[arun khatwani]

hi,

on one of the hp 9000 system we are facing issues with system time, it is not proper it is always ahead compared to its ntp server.

And to rectify the issue we do stop and start of xntpd daemon , after that the time start synchronizing and it work proper for 1 or 2 days.

And after 1 or 2 days the time again start loosing synchronization result in improper time on system

errors in syslog

/var/adm/syslog/syslog.log:May 16 01:04:53 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective hardware clock
/var/adm/syslog/syslog.log:May 16 01:21:38 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective hardware clock
/var/adm/syslog/syslog.log:May 16 01:21:57 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective hardware clock
/var/adm/syslog/syslog.log:May 16 01:38:42 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective

ive hardware clock
/var/adm/syslog/syslog.log_V1:May 15 02:53:22 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective hardware clock
/var/adm/syslog/syslog.log_V1:May 15 02:53:41 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective hardware clock
/var/adm/syslog/syslog.log_V1:May 15 03:10:26 nlxsl108 xntpd[1532]: Unsynchronized, possible remote server failure or defective hardware clock


the ntp server is pingable

Please assist to solve this

[singh sanjeev]

please do below acion and send us the output:

ps -ef | grep ntp
/usr/sbin/ntpq -p
Stop the xntpd deamon:
/sbin/init.d/xntpd stop

ntpdate -d

start the NTP deamon.
/sbin/init.d/xntpd start

[VK2COT]

Hello,

Pinging the remote server is not a good test.

ICMP simply proves that a remote server
responds to some special type of packets but
does not deal with any useful protocol like NTP.

Some of my customers completely disable ICMP
and tests like traceroute and ping are
worthless.

a) Is your remote server running NTP?

b) On your client server, run the following:

# ntpq -c peers

# ntpq -c assoc

# ntptrace -n

If your local clock is very different from a remote server, you will need to run
"ntpdate -B "...

Cheers,

VK2COT

[rick jones]

Its NTP server... any NTP client should be configured to try to get time from at least *three* NTP servers. The NTP code has logic in it which attempts to determine the suitability of the NTP servers as time sources, and that code needs three or more time sources configured.

If the upstream NTP server is not properly synchronized, it will not serve time to clients - that situation should be exposed by some of the output you've been asked to provide.