BlockAlreadyExistsException in MostAvailableFirstPolicy and RoundRobinPolicy

Description

As you can see I tried to use roundrobin and most available to make the node in cluster being balance.
I use a spark job read from current data in Alluxio and do some summary then write back to Alluxio without replace a old one.

It will be nice if you guys provide some notice about which file and folder is being missed.

Environment

Centos 6.8
Java 8
cluster include 5 nodes:

  • node1: 10GB

  • node 2,3,4,5: 20GB
    node1 also a master.

Standalone mode.
RoundRobinPolicy, MostAvailableFirstPolicy

Assignee

Unassigned

Reporter

BeanL

Labels

Components

Affects versions

Priority

Critical
Configure