Hive external table repair partition takes a long time！

[SGITLOGIN]

@lucasbak ambari version：2.7.9.0.0-26 odp version：1.2.2.0-105

Hello, my ODP cluster has been set up. Hive data is stored on oss. It takes a long time to repair the partition of the external table. Why is this? I asked this question before. You said there were no problems with your tests. Have you done any optimization on hive parameters?

Time to repair partition 271s

The number of table partitions is 31

Create table statement

[lucasbak]

@SGITLOGIN ,

Could you check the log in hive metastore while the request is being executed ?

Not at all, we have tested with HDFS (supported FS), not OSS.

[SGITLOGIN]

@lucasbak I paid attention to the log. When this sentence was printed, the SQL to repair the table partition executed in the front desk was completed.

[lucasbak]

@SGITLOGIN try to disable stats

[SGITLOGIN]

@lucasbak Please give me some guidance on how to operate it, thank you.

[lucasbak]

set hive.stats.autogather=false un hive-site.xml

[SGITLOGIN]

@lucasbak hadoop can be integrated with oss, that is, you have not tested hive data on oss, create a hive external table and msck repair tablename, right?

[SGITLOGIN]

@lucasbak I tried, changing this parameter has no effect

set hive.stats.autogather=false un hive-site.xml

[lucasbak]

@SGITLOGIN ,

indeed we have tested with Apache HDFS + Apache Hive without OSS.

[SGITLOGIN]

@lucasbak The solution has been found, just set the hive.metastore.transactional.event.listeners and hive.metastore.event.listeners parameter values ​​to empty.