[
https://issues.apache.org/jira/browse/IMPALA-7707?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16661954#comment-16661954
]
Donghui Xu edited comment on IMPALA-7707 at 10/24/18 9:26 AM:
--------------------------------------------------------------
I found the following statistics abnormal:
1、TotalNetworkSendTime is larger than 7 hours.
2、MaterializeTupleTime is larger than 40 hours.
3、ScannerThreadsVoluntaryContextSwitches is between 3k and 14k.
The contents of the hdfs directory are as follows:
partitions=1/1 files=3771 size=23.36GB
Why TotalNetworkSendTime is so large?
Is the reason for the slow query being too many query files? [~lv]
was (Author: davidxdh):
I found the following statistics abnormal:
1、TotalNetworkSendTime is larger than 7 hours.
2、MaterializeTupleTime is larger than 40 hours.
3、ScannerThreadsVoluntaryContextSwitches is between 3k and 14k.
The contents of the hdfs directory are as follows:
partitions=1/1 files=3771 size=23.36GB
Is the reason for the slow query being too many query files? [~lv]
> Impala shell takes a long time to execute insert overwrite select from
> statement
> --------------------------------------------------------------------------------
>
> Key: IMPALA-7707
> URL: https://issues.apache.org/jira/browse/IMPALA-7707
> Project: IMPALA
> Issue Type: Bug
> Components: Backend
> Affects Versions: Impala 2.5.0
> Reporter: Donghui Xu
> Priority: Major
> Attachments: profile.txt
>
>
> When impala-shell inserts a table into the partition table, it has not been
> completed for more than 7 hours.
> [~lv] Could you tell me is that a bug? The following is the profile file.
> Thanks.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
