Revision History
The v1.5.1 release has the following enhancements
- A new SNTPCLN5.NLM has been included who will load on NW 5 servers
The v1.5 release has the following enhancements
- Changed the source-address detection algorithm,
so the SourceIP parameter should no longer be needed, even on MPR
systems
- Added a thread, who checks the CMOS clock for
correct values every 60minutes
- Added AllowLargeCorrections = 1 parameter in CFG
file, so large timejumps are applied imediatly. (Not via the
tick-interface) This is only important for NW4.x
- Ticklength calculation is now more robust. When
your server shows up a lost second(s) they are (normaly) not taken into
account for the new tickrate. (This gives us a new statistic page (Key T
in the diagnostics screen))
- Better ticklength convergence The 1.4RC03 version
has the following changes compared to the previous version
- Allowed MaxTickcorrection = -1 to force initial
sync to any answer received
- Supressed Stratum display
- Corrected handling of 1 - 2 ntp servers and
UseFirstAnswer
- Ticklength is now stored as ascii text The
1.4RC02 version has the following changes compared to the previous
version
- NTPSamples added
- Up to 24 SyncAT entries allowed in the CFG file
- Corrected unload problem (abend ) when unloading
via the commandline (Open Semaphores) The 1.4RC01 version has the
following changes compared to the previous version
- ShowClock keyword to display the current time in
the upper right corner of the screen
- TickLength filter/avg. algorithm added to get
better ticklengths
- Good-server selection algorithm completely
rewritten. (Based on NTP system)
- Clock weight/avg calculation rewritten based on
NTP specifications
- Multiple log/statistic screens available
- UseFirstAnswer implemented
- The NW3.x version now converges faster to the
correct time, when negative corrections are applied
The 1.1b release has the following changes compared
to the 1.1 release:
- Fixes problems with MPR & ISWP3/4 on 4.11
servers
- Added possibility to specify source-address
(SourceIP in sntpclnt.cfg)
The 1.1 release has the following changes compared
to the 1.0 release:
- -n switch now works correctly again.
- First public release
The 1.00 release has the following changes compared
to the 0.23 release:
The 0.23 release has the following changes compared
to the 0.21 release:
- -n option works again. In the 0.2 versions, this
switch had no effect...
- Small negative corrections on 4.x servers could
result in a negative time movement for 1-2 ticks.
- Warnings added, when the 4.x server settings
could trouble sntpclnt operations.
The 0.21 release has the following
changes/enhancements compared to the 0.19 release:
This version expires 1997.5.31, a new key can be
found at http://www.neatech.ch/sntpclnt about 10 days before it expires, a
new key will be published
- More robust unload process, specialy for 3.x, but
4.x improved too
- When SYS:ETC/SNTPCLNT.CFG is found, then it is
always used, so no commandolines are needed
- No longer crashes the server, when the logfile is
open from another process
- Dynamic creation/deletion of
statistic/logscreens, based on the cfg-file
- A lot of minor enhancements to provide even
better time services
The 0.19 release has the following
changes/enhancements compared to the 0.17 release
- NDS resolution should now work correctly, so you
can specify NTP names, instead of IP numbers
- On NW4.x servers the hardware clock is now
adjusted each time we apply some changes to the systemtime.
- Configurable via a SNTPCLNT.CFG file, located in
the SYS:ETC directory (-f option)
- Start time synchronisations at a specific time.
- Possibility to do timecorrections based on the
SNTPCLNT.CFG file. So you can implement daylight saving in 3.x servers.
This feature should not be used on 4.x servers. If you need this feature
on a 4.x server, then your server is not configured correctly.
- Added activationkeys
The 0.17 version has been splitt into a NW 3.x and
NW4.x version of sntpclnt. The reason for this is, that NW4.x has a much
improved interface to the server clock. Due to this in the NW4.x version
you can get an accurancy of +-0.05 sec (NW3.x is +-0.15sec)
Beside this improved accurancy the NW4.x corrects
the tick-length when applying time corrections. (Only when the time
difference is less than the one specified with -d switch. Via the tick
interface it can take a considerable long time until the clock is finaly
set to the correct value, but it has the advantage that the clock never
runs backwards and nobody notice that the time is changing.
In this version I have added the possibility to have
a statistic screen who displays the most commonly used values. The -v
switch now creates a special screen who receives all log-informations and
it accepts now differnt levels of verbosity. Finaly with -l you can even
protocol it to a file.
Some more security checks have been made, to only
use "good" timesources, and specially solve the problem of +- 1
second differences.
This new 0.13 version of sntpclnt has two changes to
the 0.11 version. The biggest change is the implementation of a
timeresolution of about 0.08 seconds. If your internet connection is
reliable and your server is not always utilized at 100%, you will get a
time accurancy of about 0.1 seconds after 2-4 poll intervalls.
In addition we have two new commandline switches.
With -v9 the NLM displays more info on what it is doing. And with -m<secs>
you can specify the minimum time difference before a time correction is
made. Default is 0.5 sec, and you should not go below 0.1 seconds for
productional environments.
All users who like, can download the current version
to test it's correct function.
If for any chance this utility should not work on
your system, or even disturb your server (read ABEND the server) then send
me a message with as much information as possible.
I would be very glad if someone could test this
utility on 3.10 and 3.11 servers, since I only have a 4.10 and a 4.11
server available for testing.
And if you would like to see a missing feature in
the utility, then don't hesitate, and ask me. If I find it a usefull
feature too, then you have a big chance to get it implemented in a next
release.