[jira] [Created] (DRILL-4867) Scan fragment placement algorithm for parquet is not taking partition pruning into consideration

[Padma Penumarthy (JIRA)]

Padma Penumarthy created DRILL-4867:
---------------------------------------

             Summary: Scan fragment placement algorithm for parquet is not 
taking partition pruning into consideration
                 Key: DRILL-4867
                 URL: https://issues.apache.org/jira/browse/DRILL-4867
             Project: Apache Drill
          Issue Type: Bug
          Components: Execution - Flow
    Affects Versions: 1.8.0
            Reporter: Padma Penumarthy
            Assignee: Padma Penumarthy
            Priority: Critical


Drill decides how many scan fragments to run on each node based on endpoint 
affinity (ratio of number of bytes on the host/total bytes for the whole scan) 
of the node. But,  pruned rowGroups are not removed from the calculation. This 
messes up placement of scan fragment and can cause remote reads. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)