I have been trying to setup NTP on E9.2-1, which was initially failing.
However after a reboot it is now working, but in the log I get:

ntp.exe[1059]: Listening on routing socket on fd #4 for interface updates
ntp.exe[1059]: no clock adjustments will be made, DTSS is active
ntp.exe[1059]: clock adjustments will resume, DTSS no longer active

I gave up using DTSS years ago, so ayt the end of SYSTARTUP_VMS.COM I add:

$! Disable DTSS
$!
$ mcr ncl disable dtss
$ mcr ncl delete dtss

This takes a bit of time, hence the messages while it thinks about it.

I see that the DTSS startup procedures looks for a system logical
NET$DISABLE_DTSS and if set it doesn't start DTSS

If I define this in SYLOGICALS presumably it will simply no start, then
NTP can run from the off.

However I presume I will miss the automatic changes at BST begin/end,
unless NTP can be made to do that.

Any thoughts?

--
Chris

On 2023-06-05 20:17:08 +0000, Chris Townley said:

> However I presume I will miss the automatic changes at BST begin/end,
> unless NTP can be made to do that.

DTSS (DECdtss) timekeeping—including TDF and TZ and DST support and the
rest—was integrated later OpenVMS V7 and that was also then integrated
with whatever version of TCP/IP Services NTP was current back then, the
DTSS timekeeping RTL was also integrated but first documented with
OpenVMS around V8.2, and it's all managed with UTC$TIME_SETUP.COM (and
either NTP or DTSS giblets as needed), and the related database and
tooling is based on the IANA / Eggert / Olson definitions. This is all
far enough back that it's in the OpenVMS FAQ, too.

--
Pure Personal Opinion | HoffmanLabs LLC

On 05/06/2023 21:45, Stephen Hoffman wrote:
> On 2023-06-05 20:17:08 +0000, Chris Townley said:
>
>> However I presume I will miss the automatic changes at BST begin/end,
>> unless NTP can be made to do that.
>
> DTSS (DECdtss) timekeeping—including TDF and TZ and DST support and the
> rest—was integrated later OpenVMS V7 and that was also then integrated
> with whatever version of TCP/IP Services NTP was current back then, the
> DTSS timekeeping RTL was also integrated but first documented with
> OpenVMS around V8.2, and it's all managed with UTC$TIME_SETUP.COM (and
> either NTP or DTSS giblets as needed), and the related database and
> tooling is based on the IANA / Eggert / Olson definitions. This is all
> far enough back that it's in the OpenVMS FAQ, too.

Thanks, I will have a play there

--
Chris

Chris Townley schrieb am 05.06.2023 um 22:17:
> I have been trying to setup NTP on E9.2-1, which was initially failing.
> However after a reboot it is now working, but in the log I get:
>
> ntp.exe[1059]: Listening on routing socket on fd #4 for interface updates
> ntp.exe[1059]: no clock adjustments will be made, DTSS is active
> ntp.exe[1059]: clock adjustments will resume, DTSS no longer active
>
> I gave up using DTSS years ago, so ayt the end of SYSTARTUP_VMS.COM I add:
>
> $! Disable DTSS
> $!
> $ mcr ncl disable dtss
> $ mcr ncl delete dtss
>
> This takes a bit of time, hence the messages while it thinks about it.
>
> I see that the DTSS startup procedures looks for a system logical
> NET$DISABLE_DTSS and if set it doesn't start DTSS
>
> If I define this in SYLOGICALS presumably it will simply no start, then
> NTP can run from the off.
>
> However I presume I will miss the automatic changes at BST begin/end,
> unless NTP can be made to do that.
>
> Any thoughts?

NET$DISABLE_DTSS works (at least on other platforms).

Regarding DST changes, SYSGEN parameter AUTO_DLIGHT_SAV is still around
and works.

If you prefer DTSS I'm using NTP as a DTSS time provider on Alpha and
Integrity systems. SYS$EXAMPLES has a DTSS subdirectory on these
platforms containing DTSS$PROVIDER_NTP.C which I modified slightly
(iirc). Need to dig around for details and two DCL procedures to start
it at system startup. Did not try it on x86 yet.

Hope this helps,
Hans.

> [...] SYS$EXAMPLES has a DTSS subdirectory on these platforms
> containing DTSS$PROVIDER_NTP.C which I modified slightly (iirc). Need
> to dig around for details and two DCL procedures to start it at system
> startup. [...]

I don't see it on my V8.4-2L3 (IA64) system, but:

http://antinode.info/dec/sw/dtss_ntp.html

> [...] Did not try it on x86 yet.

Same here. It's been a while since I looked at the stuff.

On 2023-06-06 15:02:57 +0000, Steven Schweda said:

>>
>> [...] SYS$EXAMPLES has a DTSS subdirectory on these platforms
>> containing DTSS$PROVIDER_NTP.C which I modified slightly (iirc). Need
>> to dig around for details and two DCL procedures to start it at system
>> startup. [...]
>
> I don't see it on my V8.4-2L3 (IA64) system, but:
>
> http://antinode.info/dec/sw/dtss_ntp.html
>
>> [...] Did not try it on x86 yet.
>
> Same here. It's been a while since I looked at the stuff.

The DECnet-Plus DTSS stuff should probably be considered, um,
exceedingly relaxed, if not retired.

The particular provider file has appeared and disappeared and then
re-appeared over the years.

In the old OpenVMS FAQ, see section: 4.2 Keeping the OpenVMS system
time synchronized?

I haven't looked to see which version of NTP was in use on OpenVMS with
TCP/IP Services.

PS: DTSS$NTP_PROVIDER.C

--
Pure Personal Opinion | HoffmanLabs LLC

> PS: DTSS$NTP_PROVIDER.C

I believe that if you look hard enough, you'll be able to find both
DTSS$NTP_PROVIDER.C and DTSS$PROVIDER_NTP.C. Or else I screwed up the
names. So far as I can tell, they all date back to the 1990's, and none
of them was especially well documented.