When checking operator: A sequential operator cannot preserve the partitioning
of the parallel data set on input port 0.
, when a sequential file is connected to a transformer in a parallel extender while the transformer probagating the partition to some other stages.
But how to get rid of this warning.
Guess clearing partition is not the solution
There is a message in the FAQ forum for setting up a parallel job message handler. Because parallel jobs are chronic whingers it is almost mandatory to add a message handler to any production environment (recommend not adding one to development) to remove trivial warnings so that the job finishes with no warnings during normal operations and you don't miss more important warnings. Otherwise you have the job that cried wolf.
Hi,
I already tried to introduce message handler for this issue, but i could not convince my cleint to introdue it in their production enviroment.
My other doubt, whether the Message Id will be same across varisous versions of datastage...
Unfortunately when Ascential released the message handler functionality they did not provide any guidance as to when or why it should be used.
Here is how I use it:
No message handlers in development, unit testing, functional testing or user acceptance testing. Testers get to see the jobs run warts and all and are responsible for investigating all warning messages from all runs. Developers should try to avoid warnings.
When a job passes testing any warning messages left in the job are loaded to a message handler for use in system integration and production projects. The testers and users have accepted that those warnings are harmless. The jobs should then run in the system integration test and in production without warnings so that any new warning is found and investigated.
I've seen production jobs that occassionally dropped rows for months on end and no one investigated it because it was lost in a flood of warning messages. Messages seem to keep the same id between the 7.x releases and I see no reason why they should change.
Of course T42 once made the observation that he could make all warnings go away through good development! Have you tried putting a copy stage between the transformer and the sequential file and then playing with the partition settings of that copy stage?
Thanx for your reply vincent...
I too treat warnings seriously.....
Nope... copy stage could not help.
Need to find some better way...
I have raised the ticket in ascential, let me wait.
miwinter wrote:I have tried the copy stage theory, setting the preserve to clear, this does not suppress the warning
What is your job design, post your full warning message, addressing the the name of the stage. Most of the warning can be get ridden. Unless you have a sequential oprator connected to a preceeding operator with preserve partition flag to propagate.
Impossible doesn't mean 'it is not possible' actually means... 'NOBODY HAS DONE IT SO FAR'
