The job's interval is set to sysdate+1 results A lot of customers this time will eventually become a mess of eight bad,
or import, or manual execution, and so on, there will be automatic change, and many,
Felt somehow, and
Today, the Internet search, finally understand the reasons for the
Due to the difference in the time-checking period of Oracle's job and the number of job processes and the problem of job concurrency and job failure, the success of job deferred execution will result in a change in the start time of the next job (sysdate changes)
And also found a solution,
For example, I have to do it 1 o'clock in the morning every day , interval should be:
trunc (sysdate) +1+1/
today's morning plus day plus an hour ==> tomorrow 1 o'clock in the morning
Reason:
Sysdate will be unstable, but no matter what time of day, Trunc (sysdate) return is today's 0 points!!!!
So .... as long as the job in the day to complete can guarantee that the time point of the basic consistent!
If a half-hour interval is to be performed, the interval should be:
TRUNC (sysdate,'HH24'+ (trunc(to_char(sysdate ,'mi')/(+1)/(24 *2)
[citation reference: http://www.itpub.net/thread-264556-1-1.html]
The next execution time of Oracle database job is constantly changing.