[ https://issues.apache.org/jira/browse/HIVE-20455?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16592212#comment-16592212 ]
Daniel Dai commented on HIVE-20455: ----------------------------------- The intention is to retry ASAP when exception happen in hope it is recoverable and don't delay synchronizer. But log pollution is a bigger problem for sure. Which one is the typo, can you clarify? > Log spew from security.authorization.PrivilegeSynchonizer.run > -------------------------------------------------------------- > > Key: HIVE-20455 > URL: https://issues.apache.org/jira/browse/HIVE-20455 > Project: Hive > Issue Type: Bug > Reporter: Gopal V > Assignee: Daniel Dai > Priority: Major > Attachments: HIVE-20455.1.patch > > > Typo in the class name aside, this keeps running and thrashing through every > single DB in the cluster if not all directories are actually accessible to > the user who runs HS2. > {code} > 2018-08-24T03:14:41,932 INFO [PrivilegeSynchonizer] metastore.HiveMetaStore: > 1: get_table : tbl=hive.airline_ontime.flights_raw > 2018-08-24T03:14:41,932 INFO [PrivilegeSynchonizer] HiveMetaStore.audit: > ugi=gopal ip=unknown-ip-addr cmd=get_table : > tbl=hive.airline_ontime.flights_raw > {code} > The minimum fix needed is to move the Thread.sleep() outside of the try/catch > block, so that the sleep is triggered even if the attempt throws an exception. -- This message was sent by Atlassian JIRA (v7.6.3#76005)