running order of container within job

[haisen]

Requirement: To delete existing record in table before insert, based on a sequential file.

Version 7.2 used to have an option to delete existing records before insert but I don't see it in version 8.

Problem: Parallel job batch up the delete and insert of records. Eg. The file have 500 records, it delete 100 records based on text file, do insert for another 100 records, and then do delete & insert for another batches. Therefore, even when 500 records is expected to be loaded, less than that number is actually loaded.

Container:
As there is a few files (of different colums but same primary key positioning), we have made the following into a shared container.
same_sequential_file --------> Transformer A--------> DB (delete rows where primary key match)

In actual job (1 job only):
Shared_Container
same_sequential_file --------> Transformer B --------> DB (upsert)

There is no link between the shared container to the sequential file.
We were expecting that the container's portion will be executed followed by the upsert.

Is there anyway to set the container to complete the execution then run the upsert's flow? Must there be a link from the container to the upsert's flow?

I realised that the container can be joined to "Transformer B" then load it to DB, however only the first 3 colums of the files is same while the rest of the colums will varies.

For the same design, the container will be executed first for small volume of records.

[mahadev.v]

Without an output link from the container to the upsert flow, they will be treated as two different flows and will be executed simultaneously. And a sequential cannot have a input link and output link (unless it is a reject link). So this approach wont work. Have the second flow (upsert) in a second job. Guess that will be the best solution.