join stage production issues

mab_arif16 · Post by **mab_arif16** » Mon Jul 17, 2006 4:55 pm

HI

In some of my jobs I am facing weird problems ,I am using join stage to join the data ,Everything works fine in development env.but when I promote it to production ,the data is not joining and to the surprise not even a single row gets matched which produces almost 100,000 rows as output in development with same data .
When the project is unprotected in development then I get the required output ,but when ran in protected mode everythign gets messed up.

Are there some setting in production env. that I might be missing,or something else

Thanks
Arif

splayer · Post by **splayer** » Mon Jul 17, 2006 7:31 pm

In development, when you run it protected, do you have the same problem? By the way, what is "protected" and "unprotected"?

DSguru2B · Post by **DSguru2B** » Mon Jul 17, 2006 7:43 pm

Protected means read only. Cannot be edited.

chulett · Post by **chulett** » Mon Jul 17, 2006 8:01 pm

Actually - technically - protected is different from read only. Regardless, neither should be the cause of the issue or cause any changes in job behaviour. Unless 'protected' here means something other than 'running in a Protected project'...

kumar_s · Post by **kumar_s** » Tue Jul 18, 2006 3:09 am

Is there any change in data?
What is the data type been used?
I once faced the problem with the APT_PADCHAR property. The file produced by another job was with different padchar and hence messed up.

mab_arif16 · Post by **mab_arif16** » Tue Jul 18, 2006 9:22 am

kumar_s wrote:Is there any change in data?
What is the data type been used?
I once faced the problem with the APT_PADCHAR property. The file produced by another job was with different padchar and hence messed up.

There are some unicode feilds in one input to join on which I am joining the data ,before the join stage in a transformer I am changing them to varchar .Even in devlopment env. If I remove the unicode option even in the source before transformer I am unable to join or no output .Is there some way I can take care of this unicode .Its really giving me hard time

vcannadevula · Post by **vcannadevula** » Tue Jul 18, 2006 10:59 am

mab_arif16 wrote:
kumar_s wrote:Is there any change in data?
What is the data type been used?
I once faced the problem with the APT_PADCHAR property. The file produced by another job was with different padchar and hence messed up.
There are some unicode feilds in one input to join on which I am joining the data ,before the join stage in a transformer I am changing them to varchar .Even in devlopment env. If I remove the unicode option even in the source before transformer I am unable to join or no output .Is there some way I can take care of this unicode .Its really giving me hard time

Check the environment variables APT_NO_SORT_INSERTION/APT_NO_PARTITION_INSERTION. Depending on these values the Join stage will get effected.

kris007 · Post by **kris007** » Tue Jul 18, 2006 11:48 pm

If I remove the unicode option even in the source before transformer I am unable to join or no output .Is there some way I can take care of this unicode .Its really giving me hard time

Make sure Unicode is not there in Join Stage MetaData? But not really sure if thats causing the issue.

kool_cons · Post by **kool_cons** » Sun Jul 23, 2006 10:46 pm

Just see how many nodes are in config file..Some times if you are using multi nodes and not partitioned well , you will get this mis matched rows..In developement, may be parallel jobs running on single node.so you didn't get this error..Just check it out in production environment. I am not sure how unprotecte or protecting projec effect this issue.

kumar_s · Post by **kumar_s** » Mon Jul 24, 2006 4:35 am

Or You can tell us the difference between the Dev and PROD box, other than the procted and unprotected.