Here's the response as it appeared on vse-l this past January:

1)Setting the clock forward
There are no known CICS problems when the clock is set forward by one hour.
There is no need to shut CICS down and restart it. Once the time has been
reset on VSE, the CEMT PERFORM RESET command should be issued to
synchronise the CICS date and time of day with that of the system.
This command also adjusts the expiry time in any pending Interval Control
requests for transaction start or task delay requests.

If this CICS command is not issued, then it means that CICS will be out of
synch with the time that VSE has -- it should not cause any data integrity
problems from a CICS standpoint, but could affect time dependent
application programs. If the command is not issued, then the time held by
CICS will be one hour behind the system until either midnight or when CICS
is next restarted -- this is because CICS automatically synchronises its
time and date with that of the system at startup, and once per day at
midnight.

2)Setting the clock back
However, there is the risk of a problem when clocks are set back by one
hour if CICS is not shutdown and restarted. The risk is that if the CICS
time is reset online via the PERFORM RESET command and a CICS failure
occurs before a wrap-around of the system log has occurred, then an
emergency restart of CICS would not necessarily recover all in-flight
transactions and there would be a data integrity exposure.

The reason for this is that during its recovery processing CICS reads
backwards from the most recent record in the system log and would find a
log record time stamped more recent than the one it had just read. At this
point it would stop reading the system log, thereby not doing any backout
for in-flight transactions prior to that point.

In addition to the risk of an "unsafe" CICS restart, the log(s) containing
the disjoint time stamped records could cause problems with forward
recovery utilities that use the log(s) -- such as the CICS VSAM Recovery
(CICSVR) product and the DL/I forward recovery utilities.

There would not be any problem with CICS emergency restart if at the time
of CICS failure a wrap-around of the log had occurred, as the out of
sequence log records would have been overwritten. However, the risk would
still exist in doing any forward recovery based on any archived logs
containing records with inconsistent time-stamps.

The risk with CICS emergency restart could be avoided by doing a normal
shutdown of CICS, changing the VSE time, and restarting CICS. But again,
the risk would still exist in doing any forward recovery based on any
archived logs.

So, the only absolutely safe action when setting the clock back one hour is
to:
1) Perform a normal shutdown of CICS
2) Change the VSE system time
3) Wait one hour so that when CICS next starts the time is at least that
when it shutdown
4) Restart CICS

On Tuesday 29 October 2002 02:27 pm, Mark Perry wrote:
> Hi Rich,
> not being either a VSE or a CICS type of person, I was surprised about your
> comment "CICS journalling and warm start data tends to get a little ornery
> about the clock going backward" - is there no way to make CICS use GMT/UTC
> rather than local time for such important things as journaling?
>
> Ciao
> Mark

--
Rich Smrcina
Sytek Services, Inc.
Milwaukee, WI
[EMAIL PROTECTED]
[EMAIL PROTECTED]

Catch the WAVV!  Stay for Requirements and the Free for All!
Update your S/390 skills in 4 days for a very reasonable price.
WAVV 2003 in Winston-Salem, NC.
April 25-29, 2003
For details see http://www.wavv.org

Reply via email to