Update Query with Deadlock Error in Parallel

[SwathiCh]

Hi All,

I am trying to update 10 records in a table in oracle which are having same Key column value i.e they are having the same group value, so I want to set Y to all that records within the group.

I run the job on Default parallel node, it got aborted with the error -- Deadlock detected while waiting for the resource..

I ran the job on Single node config file, got the same error.

I added the environmental variable APT_ORA_IGNORE_CONFIG_FILE_PARALLELISM = True and run the job, got aborted with the same error.

I changed the Oracle Stage to execute in sequential mode and ran the job, it is successful.


Here what I could not understand is, running the job on single node means we are executing that in sequential manner only then why it got aborted with deadlock, and why it is successful when I convert a particular stage to run in sequential?

Any thoughts?

[swapnilverma]

.. running a job in single node does not mean it will run sequentially ... The main reason is that a node can be used multiple times in config file. ( to make logical nodes )

Running a job on single node means you are assigning resources to your job which is specific to that node....


Also if your config file shows 100 nodes does not necessarily means you have 100 actual phisical nodes ....

Even with ONE node you can achive parallelism ...

[ray.wurlod]

We need to tighten up on terminology here. For DataStage node is a single, logical, processing node. In particular, a node is not a network node name (such as occurs in the fastname property of a node definition).

[SwathiCh]

Thanks Swapnil,Ray for your responses.