Not able to load sequential file in parallel jobs

rramprasad78 · Post by **rramprasad78** » Tue Mar 25, 2008 5:48 am

Hi,

I have a .dat file which i need to load into a table. The file is fixed-width with no delimiters for fields. The file has 10 different columns and each of the columns i read them as char(n). The total length of the fixed data is 89 chars.
I am not sure if UNIX newline is the record delimiter? Though i tried with and without unix newline delimiter.
Issues that i face while trying to do View Data:
a. I am able to view the data perfectly well for the same table definition in a server job but not in a parallel job. After quite a bit of tug of war, i was able to view the data with 3 chars less in parallel job.
b. I tried a workaround for reading the record as a single entity and then do a substring. This gives a strange output: After the transformer, i inserted a Peek stage and the peek outputs the values as required ie, 89 chars displayed as per table definition. But when it writes to the output sequential file, in the last column last 2 chars are truncated though as per the definition it should have 9 chars, it displays 7 chars only.

If someone has faced an issue similar to these and have an idea of what could be the workarounds or solutions, please help me out.

Thanks in advance.

Regards,
Ram

ray.wurlod · Post by **ray.wurlod** » Tue Mar 25, 2008 5:50 am

In the table definition the Display Width is the actual width used in the file. This can be accessed as the Field Width property in extended column properties (Edit Row from the Columns grid).

rramprasad78 · Post by **rramprasad78** » Wed Mar 26, 2008 1:04 am

ray.wurlod wrote:In the table definition the Display Width is the actual width used in the file. This can be accessed as the Field Width property in extended column properties (Edit Row from the Columns grid). ...

Hi Ray,

Thanks for the suggestion. I was not aware of that.
However, my problem got resolved when i created the table definition manually. Earlier, i imported the definition from server job to parallel job. So, now i understand that could possibly be the reason. I am not a master to comment on your suggestion but the puzzle does look falling into places.
If anyone comes through this problem, do try Ray's suggestion and that should definitely save time.

Thanks.

Ramprasad