Github user srowen commented on the issue:
https://github.com/apache/spark/pull/19183
I thought this check also existed in the non-streaming code; the theory was
that if you have set a fixed number of executors but enabled dynamic
allocation, then that's probably a configuration error. But given that many
people run on clusters with dynamic allocation defaulting to 'on' globally,
that could be confusing or a little inconvenient to work around.
I don't think that check exists in the non-streaming code anymore though,
and I see a test to that effect too. Therefore I think this is reasonable for
consistency.
CC @tdas
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
