Hi guys
We are looking into parallel jobs but have met the following scenario:
Warning Limit = 1
When a warning pops up in one of the threads in the job,
this message is seen:
Issuing abort after 1 warnings logged.
but the remaining threads just keeps loading data.
At the end the job finishes with following messages:
Parallel job reports successful completion
Job completed before it could be aborted. !!!
I have warninglimit = 1 for a reason, so what is happening here?
The problem is generelly spoken a risc in all our jobs so I need one solution to cover the whole, not one solution for file stages, another for ODBC and a third for Oracle.
Anyone seen this before, and how did you manage it??
Rgds Jan
Warning in parallel job causes loss of data
Moderators: chulett, rschirm, roy
Re: Warning in parallel job causes loss of data
To me it seems like you are testing with very few data and the actual data loading is finishing even before the warning is caught by director and the job is forced to abort due to the warning limit you set.
So the questions:
1> In any of the jobs where you have already faced this situation how much data were you loading?
2> What was the warning you received in the job?
So the questions:
1> In any of the jobs where you have already faced this situation how much data were you loading?
2> What was the warning you received in the job?
It took me fifteen years to discover I had no talent for ETL, but I couldn't give it up because by that time I was too famous.
Hi Sud
Well, it may be that I use too little data, but that does not eliminate the potential of data being transferred, in cases where I want everything or nothing. Not something
The time span I see is up to 4 seconds between the warning and the completion.
My concern is that the second the warning is issued I need the remaining threads to stop , then DS can put anything in the log files it sees fit.
Number of records should be irrelevant to the fact I receive a "Job completed before it was unable to abort" !!
This makes no sense.
Actual warning is no end of record in final record of source file.
We have seen before, that DS throws a warning on PK violation also, and this is the reason we run with limit = 1.
Well, it may be that I use too little data, but that does not eliminate the potential of data being transferred, in cases where I want everything or nothing. Not something
The time span I see is up to 4 seconds between the warning and the completion.
My concern is that the second the warning is issued I need the remaining threads to stop , then DS can put anything in the log files it sees fit.
Number of records should be irrelevant to the fact I receive a "Job completed before it was unable to abort" !!
This makes no sense.
Actual warning is no end of record in final record of source file.
We have seen before, that DS throws a warning on PK violation also, and this is the reason we run with limit = 1.