Hello all,

I have a question on receiving from RM less memory/vcores than requested.

As far as I understand, when the following configurations are set, RM can
allocate more than requested(when less than the minimum):

yarn.scheduler.minimum-allocation-mb,
yarn.scheduler.minimum-allocation-vcores

I wanted to know if the reverse is possible. For example when I request a
container of 500 MB and 2 vcores, is it possible to receive a container of
400 MB and 1 core?


Thanks,
John

Reply via email to