Just to let you know....

In search of a problem completely unrelated to fpc and Lazarus I found out that the reason of the severe timing inaccuracy making this application being absolutely unusable for the desired task, has been Microsoft "Active Directory".

After I installed Automachron, syncing the PC with an "official" NTP Server, the problem vanished. (But the IT department told me that this PC will see severe issues, e.g. Active Directory preventing user login. )

I seem to remember, that many FPC built Projects either directly use the PC clock as a source for timing or use EPIKTIMER that in some incarnations use a CPU based hardware timer and in other incarnation, use the PC clock as well.

Now I found that with "Active Directory" in my installation the PC clock erratically fluctuates at rates of at least some 40 seconds per hour and at least for a total amount of 40 Seconds.

So it is a lot worse than the uninfluenced clock of even a really cheap PC and unusable for many timing tasks.

-Michael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20150608/afbc141a/attachment.html>

So, do we conclude that Microsoft's "Embrace, extend, and extinguish" strategy now applies to Time itself?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20150608/c64ba385/attachment-0001.html>

"Tony Whyman" wrote

So, do we conclude that Microsoft's "Embrace, extend, and extinguish"

strategy now applies to Time itself?

Active Directory uses Kerberos for authentication between server and client which requires the clocks of the two systems to be rather close. Now if you are in a AD the client normally uses the clock of the server, but if that fluctuates the client clock will thus fluctuate as well (we had this at my old work where the server's clock drifted heavily as well). Thus the server should normally get its time from a NTP as well...

Regards,
Sven
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20150608/43045c43/attachment.html>

Sven Barth wrote

Thus the server should normally get its time from a NTP as well...

AFAIK, the AD Server does use NTP for syncing it's own time. The clients fluctuate nonetheless.

-Michael

fluctuate nonetheless.

Then maybe the recommendation was that all systems using AD should get their time using NTP. Thus you should now have even less login problems than before :P

Regards,
Sven
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20150608/345f875e/attachment.html>

Clients using their own NTP sync is a no-go in AD, according to the IT department. You are locked in M$-Time.

-Michael

Some of this page might be interesting to you :

http://phk.freebsd.dk/time/

(tl;dr: he could detect remote NTP servers' clock PHASE fluctuation 8-O )

el es

their time using NTP. Thus you should now have even less login problems than before :P

> Clients using their own NTP sync is a no-go in AD, according to the IT

department. You are locked in M$-Time.

As long as the server and the client synchronize to the same time (not necessarily the same NTP server) in the end I see no problem. Of course it's recommended that the client synchronize their time with the AD server though as this would reduce the risk of misconfiguration.

Regards,
Sven
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20150609/692700c5/attachment-0001.html>