xntp issues

[nyck]

Hello,

When configuring AIX5.3 with regard to xntp I’m getting the following error message after the xntpd bombs out:-

---------------------------------------------------------------------------
LABEL: SRC_SVKO
IDENTIFIER: BC3BE5A3

Date/Time: Fri 14 Mar 10:43:58 2008
Sequence Number: 18
Machine Id: 000370DAD300
Node Id: lonibm8
Class: S
Type: PERM
Resource Name: SRC

Description
SOFTWARE PROGRAM ERROR

Probable Causes
APPLICATION PROGRAM

Failure Causes
SOFTWARE PROGRAM

Recommended Actions
MANUALLY RESTART SUBSYSTEM IF NEEDED

Detail Data
SYMPTOM CODE
256
SOFTWARE ERROR CODE
-9017
ERROR CODE
0
DETECTING MODULE
'srchevn.c'@line:'350'
FAILING MODULE
Xntpd

Any ideas?

I used the same process for this that worked a few months ago on my other IBM servers. The process I’m using is as follows:-

Uncomment xntpd in /etc/rc.tcpip
Add the entries for the NTP server into /etc/ntp.conf
ntpdate <ip_address_of_NTP_server>
startsrc –s xntpd

xntpd starts up and after about 5 mins it crashes out, most annoying!

 

[Breslau]

what type of time server are you using, well known internet, or a local one?

 

[nyck]

My NTP server is a Sun V890 which points to server elsewhere in our company. The issue is that this only happens with my new IBM P505 servers currently. I just enabled ntp on my new Itanium servers and no issues what so ever. One one of the problem IBM servers I set the date to be as close as possible to the NTP server manually and then started xntp. When I run an ntpq>pe the offset is:-

3600356

this is way too high!

 

[Breslau]

hmm, if you have a big difference to make up between your ntp server and an AIX client, you can put this in your ntp.conf file:

ignore_bigtimestep yes

the client will match up with the server rather quickly. some folks choose not to have this feature because it can dramatically change time on a host over a short period. for me it is not a problem.

one think you can verify is that your AIX clients have the correct time zone configured, and that they are speaking the same version(s) as your SUN box. although in most cases, i'd think this shouldn't be a concern.

 

[nyck]

I have just configured three AIX clients with xntpd and they have all worked fine. When I restart xntpd and ran ntpq>pe the offset value had dropped dramatically for no apparant reason, maybe there was a network glitch of something like that!

The one issue I noticed was that the time had sync'd with the NTP server but the clock was an hour ahead. So on the NTP server I ran an echo $TZ and it came back with GB. So on the AIX clients I changed the TZ variable in /etc/enviroment to be GB, logged out and back in again and all was sync'd up perfectly.

The ignore_bigtimestep variable does it go in the client ntp.conf or the server ntp.conf?

 

[Breslau]

that would go into the client's config file. it tells the daemon to ignore a big change in the time differential with a server.

 

[p5wizard]

sounds like it was a TZ problem, nothing more.

difference of 3600356 is really a difference of 1h (3600s or 3600000ms - TZ difference between server and client) + 356ms (real difference between server and client).

HTH,

p5wizard