[
https://issues.apache.org/jira/browse/IMPALA-12067?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhi Tang updated IMPALA-12067:
------------------------------
Description: Although catalogd propagates incremental metadata updates at
the partition level, the coordinators still loads all of the table's partition
metadata at query time, even if it is not being used. This is caused by coor
needing to calculate the total hdFsbytes of the HDFS table. It can be optimized
by transferring the computational logic to catalogd. (was: Although catalogd
propagates incremental metadata updates at the partition level, the
coordinators still loads all of the table's partition metadata at query time,
even if it is not being used. This is caused by coor needing to calculate the
total hdFsbytes of the HDFS table. This can be optimized by transferring the
computational logic to catalogd.)
> Avoid coordinator loading useless HDFS table’s partition metadata.
> ------------------------------------------------------------------
>
> Key: IMPALA-12067
> URL: https://issues.apache.org/jira/browse/IMPALA-12067
> Project: IMPALA
> Issue Type: Bug
> Reporter: Zhi Tang
> Priority: Major
>
> Although catalogd propagates incremental metadata updates at the partition
> level, the coordinators still loads all of the table's partition metadata at
> query time, even if it is not being used. This is caused by coor needing to
> calculate the total hdFsbytes of the HDFS table. It can be optimized by
> transferring the computational logic to catalogd.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
