time server

[roberto salvatori]

i need to setup a time server.
this is my situation:
i have 3 hp-ux 11 server and one of these must be a time server. but the problem is that these servers cannot see the reality outside my network. what can i do?

[Ian Dennison_1]

http://www.leunig.de/en/produkte/netzwerk/funkuhr/index.shtml

Set up an NT Server with the above mentioned product (its read only, should have no output), and then set up a NTP domain within your UNIX Servers, with one primary and 2 alternates. They should sync between each other, and to the Atomic Clock product above.

Share and Enjoy! Ian

[Jannik]

If they have no way of getting the info from the outside then you could:
1 - setup one of the servers as ntp server.
2 - set all others up as clients

Then you would of cause have to check the time on the server, but the you only have to check one place. And all would be the same amount of time out sync :-)

Another solution would be to connect one or more of the servers to a GPRS (i think) that could aquire the information.

Best regards.

[Caesar_3]

Hello!

For the config you can find info on docs.hp.com
And for the wright time on the server if you
don't have connection to the world you can do:
1. Every day check the time and if need so set the wright.
2. To buy GPS and connect to server.
3. To buy Atomic clock and connect to server.

Caesar

[roberto salvatori]

thanks for your faster answer but i don't know to setup ntp server. can you help me?

[Ian Dennison_1]

Commands as follows,...

sam - Time - NTP Network Time Sources

Actions - Add remote server or peer - (Enter IP Address), select Server, OK.

NB Set up all 3 servers as Peers and they should correct themselves. Alternatively, set up the Atomic Clock mentioned above, have 1 server sync to that, and all other sync to the server.

There is some excellent help in the NT Network Time Sources, under the option "Actions" - "Explain NTP".

Share and Enjoy! Ian

[T G Manikandan]

check this doc

http://www1.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000064722996

[roberto salvatori]

hi Manikandam (i hope that i wrote your name well), i've follow your suggestion and i made all steps contained in a document but no way to update a time on a clients.
i've checked that a ntp daemons was running and all ok. it seems that the clients didn't seen broadcast time. what's wrong?

[Robert-Jan Goossens · Accepted solution]

Hi Roberto,

getting war in Italie ? (:-)

Have You got an other ntp-server in your network NT,WIN2000 etc ?

I use a gps-device as a primary ntp server - a hpux server as secondary ntp server for the unix servers, if you like I mail the conf of the hpux server. But with sam it is realy easy.

Kind regards,

Robert-Jan.

[Robert-Jan Goossens]

Ooops war should be warmmmmmmmmmmm with a mmmmm

big typo (:-)

[James R. Ferguson]

Hi Roberto:

One common problem when setting up NTP services is a time difference greater than 1000-seconds. Differences this large or larger are rejected to prevent insanity. The 'xntpd' daemon will eventually die if it cannot find a source to which to synchronize.

If your server's differ by this amount of time or more, you will need to adjust their time with the 'date' command before you start 'xntpd'. If you are setting their clock backwards, be sure tio stop any databases or time-dependent applications first. Then reset the clock and reboot for a clean startup. If you are setting the clock forward, you can generally skip the reboot.

Consider, too, using the 'ntpq' utility to diagnose your inability to synchronize time services. This utility, along with other troubleshooting tips is discussed in chapter-7 of "Installing and Administering Internet Services":

http://docs.hp.com/hpux/onlinedocs/B2355-90685/B2355-90685.html

Regards!

...JRF...

[T G Manikandan]

Client configuration

Make sure you have the server name in the /etc/ntp.conf file

like

server

change

XNTPD=1
in the /etc/rc.config.d/netdaemons file

start the xntpd using

#/sbin/init.d/xntpd start
Then do a
#ntpq -p
to check that the client can see the server

Revert

[roberto salvatori]

first of all thanks all. i really appreciate your help.
ok, the situation is strange cause i think all is ok.
i checked on ntp server the configurazion and the ntp.conf contains:

server itdevux version 3 prefer
server 127.127.1.1 minipoll 3 maxpoll 4
broadcast 128.101.255.255 version 3

the /etc/rc.config.d/netdaemons contains:

export NTPDATE_SERVER=
XNTPD=1
export XNTPD_ARGS

i stopped and started the ntp daemon many and many times.
i made ntpq -p itdevux (from a client ntp) and the server is up and running.
where's the problem then?

[roberto salvatori]

i forgot to say no way to syncronize time on servers.
thanks

for Robert-Jan: i don't understand what u mean with warm, eheheheeh. if u want to send me your ntp.conf file i will be very happy. so i can check with my files (if u want don't forget to send me server file and client file)

[T G Manikandan]

First I would like to know your setup first.
Briefly describe!
Also,
Any xntpd messages in /var/adm/syslog/syslog.log file.

[Robert-Jan Goossens]

Roberto,

I meant the heat you got in Italie, 37 or 38 degrees celcius. (:-)

0 points

[Bill Douglass]

Try ntpq -p with no argument. It should show you the ntp server name with a * in front of it. If not, then your clients are not syncing to the host. If you see another server listed here, then you have more than one ntp server on your network.

What is the time difference between your clients and the ntp server

ntpdate -q servername

If the offset is over 1000 seconds, then ntpd will fail. You can initially set the time with

ntpdate servername

and then start ntpd.

[Joris Denayer]

Roberto,

Can you post the output of the command
# ntpq -p
from the server as well as the client(s)

Joris

[roberto salvatori]

yes, i can all that u want to know.
what i did to configure a ntp server and client are the same steps described on this document:
http://www4.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&admit=-938907319+1056708501152+28353475&docId=200000064722996

this is all what u need.

************* on client machine *************

itlabtux:root: /usr/bin# ntpq -p 128.101.0.19
remote refid st t when poll reach delay offset disp
==============================================================================
LOCAL(1) LOCAL(1) 3 l 116 64 7 0.00 0.000 3885.01
128.101.255.255 0.0.0.0 16 - - 64 0 0.00 0.000 16000.0
itdevux 0.0.0.0 16 - - 64 0 0.00 0.000 16000.0

itlabtux:root: /usr/bin# ntpq -p 128.101.0.19
remote refid st t when poll reach delay offset disp
==============================================================================
LOCAL(1) LOCAL(1) 3 l 120 64 17 0.00 0.000 1885.01
128.101.255.255 0.0.0.0 16 - - 64 0 0.00 0.000 16000.0
itdevux 0.0.0.0 16 - - 64 0 0.00 0.000 16000.0

itlabtux:root: /usr/bin# ntpdate -q 128.101.0.19
server 128.101.0.19, stratum 16, offset -70.413622, delay 0.02580
27 Jun 14:16:16 ntpdate[24549]: no server suitable for synchronization found

itlabtux:root: /usr/bin# ntpq -p
No association ID's returned

no messages on syslog.log


************* on ntp server machine *************

after leaved minpoll and maxpool (it was wrong) and restrted the daemon
/var/adm/syslog/syslog.log
Jun 27 14:16:53 itdevux xntpd[6484]: xntpd version 3.5f: Tue Jun 25 12:40:32 IST 2002 PHNE_27223
Jun 27 14:16:53 itdevux xntpd[6484]: tickadj = 625, tick = 10000, tvu_maxslew = 61875
Jun 27 14:16:53 itdevux xntpd[6484]: precision = 6 usec

itdevux:root: /# ntpq -p
remote refid st t when poll reach delay offset disp
==============================================================================
*LOCAL(1) LOCAL(1) 3 l 18 64 37 0.00 0.000 885.01
128.101.255.255 0.0.0.0 16 - - 64 0 0.00 0.000 16000.0
itdevux 0.0.0.0 16 - - 64 0 0.00 0.000 16000.0

i will waiting for you.

ever for Robert-Jan: yes, so so hot here. and we have also electric problem concerning high use of electricity. it seems to stay in Hollywood but here there is not a ocean.

[Joris Denayer]

Hi,

itlabtux:root: /usr/bin# ntpq -p
No association ID's returned
|
V
I think that xntpd is not running on your client. Check with the ps-command.



If you start xntpd on the server, it will not answer ntp-requests until it is sync'ed with it's LOCAL Clock (127.127.1.1). When it is synced, you'll observe the (*) on the first column of the ntpq -p output.
This occurs only when the reachability flags are 37 or higher, this means ~ 64*5 sec = ~ 5'20"

Hope this helps