Re: Will time synchronization affect Oracle?

Years ago, I used to work with ADABAS database systems. On those systems, every transaction is time-stamped. When setting the clock backwards, we had to leave the database down for one hour so that we would not be creating any new transactions with younger timestamps than previous transactions. Since Oracle uses a SCN for identifying transactions, one would think that setting the time back would not necessarily be a bad thing. And that may be true. But many systems do query SYSDATE and store the value somewhere. For daylight savings time changes in the fall, I always like to leave the databases down for an hour. Probably just my paranoia from my ADABAS days.

That being said, in Y2K testing, we made sure we had a good backup of our database. We issued time changes without restarting the database. We didn't have any db crashes. After testing, we just restored our database from the backup so that no "inaccurate" timestamps remained in the db.

HTH,
Brian

Jeff Hunter wrote:
>
> Something I found in Y2K testing was that setting the date either backwards
> or forward on 2.6 via the "date" command will most likely crash your
> database. When there is little activity going on, sometimes it works
> successfully, but most times, no.
>
> We have all our servers' times co-ordinated via NTP and it works very
> well...
>
> Chuck <chuck_at_phi.org> wrote in message
> news:38722eca.71969636_at_client.nw.news.psi.net...
> >
> >
> > I have some 805 db's running on Sol. 2.6. I sync the times of these
> > servers frequently using rdate. The db's are not running in archive
> > log mode. How does Oracle respond to changes in the system time?
> >
> >
> > Thanks,
> > CHuck
Received on Thu Jan 06 2000 - 09:11:49 CST