[REEF-1688] IMRU driver fails if RM allocates extra evaluators

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.16
• Component/s: IMRU
• Assignee: Mariia Mykhailova
• Reporter: Andrey
• Created: 15/Dec/16 18:43
• Updated: 05/Jan/17 01:45
• Resolved: 19/Dec/16 18:13

Description

if Yarn allocates more evaluators than IMRU driver requested, IMRU driver fails.
Seems like RM is not behaving well in this situation, but IMRU driver could continue working with existing evaluators and just reject the extra allocations.
Would make it more robust.

The error we log and fail with:
"Possible Bug in the IMRU code: Trying to add an additional Evaluator container_####, but the total expected Evaluator number <RequestedN> has been reached.

Issue Links

• is related to: REEF-1692 Revert the ignorance for extra Evaluators requested