When we store data, the datanode stores the metadata that has the location of the blocks of the files that we have stored.
But I wonder, if we run a task with Yarn, he looks for the hardware resources to execute that task.
But at some point, I understand that it has to communicate with the NameNode to know in which nodes are located the blocks that make up the files required by the task to be executed.
Can someone please clarify this?
Greetings