hello, everyone!
this post i want to introduce you a change faults of HiveServer startup failure due to Hive Metadata upgrade timeout. let's see it together.
[Applicable Version]
C70SPC2xx-6.5.1.6
C70SPC2xx to 6.5.1.7 (This case uses the upgrade from C70SPC200 to 6.5.1.6 as an example.)
[Context and Symptom]
After C70SPC2 is upgraded to 6.5.1.6 or 6.5.1.7, the HiveServer fails to be started and the message "Could not connect to meta store using any of the URIs porvided." is displayed.
The message "Hive Schema version 3.1.0 does not match metastore's schema version 1.3.0 Metastore is not upgraded ot corrupt" is displayed in the MetaStore log.
[Cause Analysis]
During the upgrade, the pre-upgrade check is not performed according to the upgrade guide. As a result, the timeout interval of the Hive upgrade metadata operation is not checked.
Modify the socketTimeout value of javax.jdo.option.ConnectionURL in the ${BIGDATA_HOME}/FusionInsight_HD_6.5.1/*_*_MetaStore/etc/hivemetastore-site.xml file to 1800 seconds.
Modify the socketTimeout value of javax.jdo.option.ConnectionURL in the ${BIGDATA_HOME}/FusionInsight_HD_6.5.1.7/*_*_MetaStore/etc/hivemetastore-site.xml file to 1800 seconds.
[Solution]
Use PuTTY to log in to the MetaStore node as user root.
Run the following command to disable logout on timeout:
TMOUT=0
Run the following command to perform a query:
find /opt/huawei/Bigdata -name upgrade-1.3.0-to-3.1.0.postgres.sql
The content of the following two SQL files is the same. The file contains 577 lines. The SQL statement containing 796 lines is for the rolling upgrade.
Log in to the hivemeta database of DBService as user omm and run the following script (you are advised to manually copy the SQL script and run it by line):
Restart the HiveServer.
