Details
-
Type:
Bug
-
Status: Resolved
-
Priority:
Minor
-
Resolution: Won't Fix
-
Affects Version/s: CDH3b4
-
Fix Version/s: None
-
Component/s: Hive
-
Labels:
-
Environment:CentOS 5.2, 10 node cluster, hive on a separate client machine.
Description
After running the upgrade scripts under /var/lib/hive/scripts/metastore/upgrade/mysql/ against our production MySQL database containing our hive metastore, we noticed a significant slowdown between job submission to the cluster, and actual job execution (even simply appearing on the JobTracker wasn't immediate).
We tried several different things, reviewing DNS issues, modifying /etc/hosts, reviewing datanode, tasktracker, jobtracker, and namenode log files in DEBUG mode, etc.
In the end, removing hive database from MySQL, and letting hive recreate the schema from scratch resolved the issue.