[Question][Worker] The worker should perform a system load check before executing the task.

[1105560808]

Search before asking

• I had searched in the issues and found no similar feature requirement.

Description

I found that in the current 3.2.2 release version, the worker checks whether a task can be executed only by determining if the WorkerTaskExecutorThreadPool is full. However, in some cases where tasks consume more resources, it is possible that the memory and CPU are already fully utilized even though the WorkerTaskExecutorThreadPool is not full. Continuing to execute tasks in such cases could cause worker task execution to fail. Furthermore, when the master uses the LowerWeight distribution strategy, it selects the node with the lowest weight based on metrics for task distribution. Shouldn't there be an additional check to either reject the task distribution or place it in a queue to wait for the next distribution? For example, when all worker nodes' CPU usage reaches 100% or 90% or more, there would be no need to distribute tasks further.

Are you willing to submit a PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[SbloodyS]

This has been implemented, please consult the documentation.