Node And partiioning clarification

[myukassign]

I need a help to get my concept validated... I read many articles and I have a mixed understanding about partitioning and node.

Q1. If I have 4 nodes configured in my config file. Any parallel job that running in my server will make exactly 4 copies of my job. Is that right?

Q2. If I hash partitioned my stages and I have only 2 records coming. At the max it will process the data through two nodes and the remaining copies of job in the other two nodes will become void.

Please correct me if my understanding is wrong gurus. Many thanks for your time reading and replying in advance.

[tbharathkumar]

Q1: Yes, Four instances will create
Q2:Yes, I guess you are correct

[ray.wurlod]

1. Assuming that four nodes are in your default node pool and no node pools are constraining which nodes stages execute on and provided no stage executes in sequential mode, yes, stages will execute using four nodes. But this is not the same as "four copies of your job". One copy (the "score") is made, and distributed to four processing nodes.

2. At least two nodes will process zero rows. There is no concept of "become void". The two rows may be executed on two nodes or may be executed on one node, depending on the partitioning algorithm, node pools and whether the stages are executing in sequential mode.