On Tue, 2 Nov 2021 11:52:21 -0500, Paul Gilmartin <[email protected]> wrote:

>z/OS shirks the issue by making user address spaces non-dispatchable during a 
>leap second.

I wouldn't say "shirks".  Perhaps "accommodates" might be a better word. 

If you don't configure leap seconds in the timing network defintition (CTN), 
STP thinks Coordinated Server Time (CST) is wrong and begins to steer out the 
1-second difference.  As Tim noted, that takes about 7 hours.  Those with 
strict time stamp requirements can't tolerate it being wrong for 7 hours, so 
suspending the application for one second is preferable.

> z/VM?

Tim pointed to my article.  z/VM will give the wrong time if you have leap 
seconds configured in the CTN.

>I have seen some discussion of inserting leap seconds at 23:59:60 local
>time rather than UTC.  Bad Idea.

The discussion getting more traction in the scientific community is to abandon 
the leap second entirely.

Alan Altmark
IBM

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to