NTP client on VMS can't talk to NTP Server

[Ros Archer]

We have a stand-alone network consisting of a GPS/NTP Server (NTPv4), Windows Server, Alpha Server (OpenVMS 7.1-2, NTPv3, ntp client) and Sun Server (ntp client). Currently the Windows Server can synchronise with the NTP Server. I haven't even started on the Sun Server, I'm just concentrating on getting the Alpha to sync with the NTP Server.

On the Alpha, I have enabled and started the NTP Server via the @sys$startup:tcpip$config tool.

I have a tcpip$ntp.conf file that resides in sys$sysroot:[tcpip$ntp] directory. This is a copy of the template file, except that the only original line left un-commented is for the drift file. I've added a line for the NTP Server:
server xxx.xxx.xxx.2
and a line for additional logging
logconfig +clockall +peerall +sysall +syncall

To kick it off, I use NTPDATE:
@sys$startup:tcpip$ntp_shutdown
@sys$manager:tcpip$define_commands
ntpdate -d xxx.xxx.xxx.2

The response looks like it is populated by zeros:
server xxx.xxx.xxx.2 (xxx.xxx.xxx.2), port 123
startum 0, precision 0, leap 00
refid [0] delay +0.00 dispersion +64.00 offset +0.00
transmitted 4, in filter 4
reference time: 17-NOV-1858 00:00:00.00
originate timestamp: 17-NOV-1858 00:00:00.00
transmit timestamp: 6-SEP-2005 11:06:23.52
filter delay: +0.00 ...
filter offset: +0.00 ...
No server suitable for synchronization found from those provided.

To startup NTP and view the peers list, I use:
@sys$startup:tcpip$ntp_startup
ntpdc
peers

I get the following output:
remote local st poll ==============================================
=xxx.xxx.xxx.2 xxx.xxx.xxx.140 16 64

reach delay offset disp
===========================
0 0.00 0.00 16.0

The stratum of the .2 NTP Server is too low for the Alpha to sync to, and I'm guessing that it's because the time between the originate timestamp and the transmit timestamp is more than 1000 seconds (140 odd years more).

On looking through the log file, the following appears:
receive from xxx.xxx.xxx.2
receive: his mode 4
poll_update(xxx.xxx.xxx.2, 6, 1)
invalid packet header xxx.xxx.xxx.2 0x20 0.000 00.0012221

I read somewhere that 0x20 is the error number associated with a stratum level being too low to sync to.

Any ideas what's going wrong here? I believe that an NTP Server of v4 can sync an NTP Client of v3.

The confusing part is that it is working perfectly on the windows server - I can't blame the clock!

[Ian Miller. · Accepted solution]

I wonder if your server is talking sntp which windows systems understand but is not proper ntp. The current version of ntp is V3 as V4 is not approved yet.

See
http://www.eecis.udel.edu/~ntp/ntpfaq/NTP-s-def.htm

[Antoniov.]

Hi Ros,
nice to meet you and welcome to VMS forum.

At first time, I guess openVms use NTP V4 because in my NTP log I read ntpd V4.1 (see also ntpdate -h). At moment I don't remember standard version of various NTP.

I did'nt understand what did it happen; may you attach TCPIP$NTP_RUN.LOG ?

Antonio Vigliotti

[Ian Miller.]

ntpd V4.1 is the version of the software.
I was referring to the version of the protocol.

[Ros Archer]

Thank you both for your prompt responses!

The NTP server could be running SNTP - the documentation doesn't say, but I'm suspicious. I've written to the manufacturers to find out...

The windows server is capable of passing on the time to its clients using NTP as opposed to SNTP. So I've added the windows server to the NTP.CONF file as another "server" line and it works! I don't want the alpha to update the windows server, so I haven't made them peers.

Thanks for the suggestions - this forum is great!

[Ros Archer]

Alpha now points to Windows Server for NTP.

[Antoniov.]

Ian,
just for talking, your link refers to VMS V7.2 with NTP V3. Now, with TCP/IP V5.4 on V7.3, NTP version is V4.
See official HP guide http://h71000.www7.hp.com/doc/732final/6526/6526pro_028.html#ntp_chap
for NTP service in TCP/IP V5.4.

Antonio Vigliotti