[SAMZA-1197] Locality should be used in standalone execution (host-affinity)

• Type: Sub-task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Fix Version/s: 0.15.0
• Assignee: Navina Ramesh
• Reporter: Navina Ramesh
• Created: 07/Apr/17 21:45
• Updated: 12/Jan/18 19:26

Description

With SAMZA-1084 and SEP-1, we started using a varying list of processorId (String), instead of a monotonically increasing bounded set for integers as containerId. This is essential to leverage data-locality for processing, aka host-affinity feature.
Some of the changes related to making this work:
1. Use localitymanager in JobModel Manager
2.GroupByContainerCount (which is the default TaskNameGrouper) assumes the bounded integer of containerId for rebalancing tasks. This should be changes to work with a varying set of processorId Strings.