Riza Suminto created IMPALA-13997:
-------------------------------------

             Summary: Mention num_host seen in non-OK status of 
Scheduler::CheckEffectiveInstanceCount
                 Key: IMPALA-13997
                 URL: https://issues.apache.org/jira/browse/IMPALA-13997
             Project: IMPALA
          Issue Type: Improvement
          Components: Distributed Exec
            Reporter: Riza Suminto


There is a sanity check in Scheduler::CheckEffectiveInstanceCount if 
largest_inst_per_host > qc.MAX_FRAGMENT_INSTANCES_PER_NODE

[https://github.com/apache/impala/blob/ef8f8ca27b52f7fd842a7a887d5c9a8db9831f79/be/src/scheduling/scheduler.cc#L457-L462]
 

We should print num_host seen by scheduler or any other executor group set / 
request pool information that can be helpful in triaging such issue.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to