DSXchange

Yes a sequencer,
read the parameter from unix file before the job and loop the job.

YES.
you need to elaborate and be specific on your requirement.

set parameters
$APT_IMPORT_PATTERN_USES_FILESET = True
$APT_IMPORT_PATTERN_USES_FILESET_MOUNTED = True

can you give more details?
how is some lines in the 090309.sdi are affected to the 090311.sdi file?

some object for this job is corrupted, so you need to delete it from repository tables.

1. Get the Job id
2. delete from ds_jobs where NAME =<job name>
3. delete from JOB_OBJECTS WHERE NAME =<job name>

just check the column names, not sure its NAME or JOBNAME.

that was just an idea.
you can have a condition , if second file size is greater then Zero then only append. Else delete the first file and then touch <file name>

OK , Create two files in the job, one file for only header records with column names only, another file with records only and then in after job subroutine, append the second file into first file.

please try to be very clear on your requirement.
do you mean if there are no record then empty file should be generated else the first row should be the column names??

this has been discussed so many times... just try some search "capture duplicate"

did you try the search

i find this viewtopic.php?t=110621

if i can call the script in a datastage stage i think it will run on multiple nodes

and why do you think so???

it will run on head node only

when you re execute the job, a new ORCH_WORK_<hexvalue> table will be created. and this is probably because the <hexvalue> is not stored anywhere.
unlike the MLOAD stage there is no option to define ORCH_WORK and ORCH_ERR table in teradata enterprise stage.

do you know why the upsert mode tries to proceed updates in a empty table with unique keys in the source ??

after certain point, its not an empty table, there is a parameter for it to set after how many records it will be committed.

Use any third party tool like Control-M

what is your findings?
are you able to load or not?