I am trying to understand fload from a restrating point of view.
According to the fast load manual there are 2 phases in it
loading phase and end loading phase and the restart strategy depends on the phase in which it got aborted.
Could any one explain what is achieved in these two and how do we restart a script which fails in either of them?
How do we identify in which phase the job has failed by looking at the logs ?
According to my understanding
loading phase - corresponds to the begin loading staement and puts data into spool (work tables)
end loading phase - corresponds to the end loading statement sorts and loads the target table
Its kind of clear that when it fails in loading phase then we have to resubmit the script without any of the delete,drop or create statements.
Also the manual says that if a Teradata FastLoad job was paused during the end loading phase then to restart we just need to submit BEGIN LOADING and END LOADING commands alone
In this case will it load all the data or just release the lock on target table ?
How can a job get aborted during the end loading phase can we simulate it by pressing CTRL C 3 times ?
Looking for valuable inputs
Please refer to FastLoad Manual Chapter 3 End Loading Command. It mentions:
"The END LOADING command begins the end loading phase of a Teradata FastLoad job. During this phase, all rows are distributed on the AMPs and stored in the final Teradata FastLoad table.
The Teradata Database uses internal checkpointing while processing an END LOADING command. Therefore, a job can be interrupted during the end loading phase without disturbing processing status. When the job is restarted, the Teradata Database resumes processing from where it left off."
After the end loading phase restart completes, the Teradata Database removes the access locks that were placed on the three tables specified in the BEGIN LOADING command.
Then FastLoad deletes error tables if they are empty and restart info from the log table.