Configuring NTP

[Manny DeAssis]

Attempted to configure NTP on OpenVMS V7.3-2/1 systems running TCPIP V5.4 ECO 2 / V5.3 EC0 2. When using TCPIP$CONFIG to Enable and Start NTP service I get the following error:

%TCPIP-I-INFO, image SYS$SYSTEM:TCPIP$NTP.EXE installed
%TCPIP-I-INFO, logical names created
%TCPIP-E-STARTERROR, error starting NTP service
-TCPIP-E-BADVALUE, value Service User_name is incorrect
-RMS-E-RNF, record not found
%TCPIP-E-STARTERROR, error starting NTP service
-RMS-E-RNF, record not found
%TCPIP-E-STARTERROR, error starting NTP service
-TCPIP-E-BADVALUE, value Service User_name is incorrect
-RMS-E-RNF, record not found
%TCPIP-E-STARTERROR, error starting NTP service
-RMS-E-RNF, record not found
%TCPIP-E-STARTFAIL, failed to start TCPIP$NTP
-TCPIP-E-ENABLERR, error enabling service

This is a nine node cluster. So far I've been successfull at configuring and enabling on only two systems. They all have basically the same copy of tcpip$ntp.conf. Each of them are configured as NTP clients. Any suggestions?

[Garry Fruth]

Does the username TCPIP$NTP exist on each node?

[Karl Rohwedder]

Did you configure NTP using TCPIP$CONFIG.COM?
This will check and create the TCPIP$NTP user account. Did all cluster nodes share a common
SYSUAF?

[Wim Van den Wyngaert]

Googled it : could it be

There was product installed (DCE) that used its own NTP and
has its own service configured to use port 123.

Wim

[faris_3]

Hi,

How many system disks ?
How many sysuaf.dat ?

/hf

[Manny DeAssis]

This is nine node cluster. Two UAFs. One shared by three nodes (Dev). The other shared by six (QA). NTP was configured via TCPIP$CONFIG procedure. TCPIP$NTP account exists in both UAFs. DCE is installed and running on two of the nine. They share the same UAF (QA) so they're in the larger group. I tried removing the NTP service and re-installing I still cannot start it on seven of the nine. That's right, only two have been successful with starting NTP. These are not the two with DCE.

[Wim Van den Wyngaert]

May be something else used port 123 ?
ucx show dev may help.

Wim

[Bojan Nemec]

Manny,

Try to dump the sys$system:tcpip$service.dat and look if there is a correct username in the ntp service:

$ DUMP/PAGE/RECORD SYS$SYSTEM:TCPIP$SERVICE.DAT

At the beginning of the record you have the service name (first dump line) and then you have the service username at the 28th byte (second dump line last 4 characters and third dump line). Service name and username are filled with blancs.
If the username is not proper (TCPIP$NTP) you can change it with:

TCPIP> SET SERVICE NTP/USERNAME=TCPIP$NTP

Bojan

[Manny DeAssis]

I got HP involved with this one. After about 25 minutes of trouble-shooting the problem presented itself in the form of a corrupt TCPIP$SERVICE.DAT file. I was able to get NTP to start only after copying a known good TCPIP$SERVICE.DAT file from one of the two systems where NTP was working.

Thanks.

[Wim Van den Wyngaert]

Do you know why it got corrupted ?

Wim