[ 
https://issues.apache.org/jira/browse/HIVE-26740?focusedWorklogId=826170&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-826170
 ]

ASF GitHub Bot logged work on HIVE-26740:
-----------------------------------------

                Author: ASF GitHub Bot
            Created on: 15/Nov/22 14:32
            Start Date: 15/Nov/22 14:32
    Worklog Time Spent: 10m 
      Work Description: szlta opened a new pull request, #3765:
URL: https://github.com/apache/hive/pull/3765

   HIVE-20172 solved an issue where HMS was running the compaction worker, and 
StatUpdater thus ran in the very same JVM causing kerberos issues while 
connecting to "itself". The fix was to update HMS URI setting to empty string. 
This way HMS wouldn't have to go around itself to connect to itself :)
   
   But.. this is not correct anymore, as Hive doesn't support running the 
compaction worker on HMS side anymore, but only on HS2. So in any case this 
code part ends up executed in HS2, where an embedded HMS will be created due to 
this empty string URIs config, and eventually will open connections directly to 
HMS backend DB.
   
   We need to remove this setting as it was fixing something that is not 
relevant anymore.
   




Issue Time Tracking
-------------------

            Worklog Id:     (was: 826170)
    Remaining Estimate: 0h
            Time Spent: 10m

> HS2 makes direct connections to HMS backend DB due to Compaction/StatsUpdater
> -----------------------------------------------------------------------------
>
>                 Key: HIVE-26740
>                 URL: https://issues.apache.org/jira/browse/HIVE-26740
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Ádám Szita
>            Assignee: Ádám Szita
>            Priority: Major
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> HIVE-20172 solved an issue where HMS was running the compaction worker, and 
> StatUpdater thus ran in the very same JVM causing kerberos issues while 
> connecting to "itself". The fix was to update HMS URI setting to empty 
> string. This way HMS wouldn't have to go around itself to connect to itself :)
> But.. this is not correct anymore, as Hive doesn't support running the 
> compaction worker on HMS side anymore, but only on HS2. So in any case this 
> code part ends up executed in HS2, where an embedded HMS will be created due 
> to this empty string URIs config, and eventually will open connections 
> directly to HMS backend DB.
> We need to remove this setting as it was fixing something that is not 
> relevant anymore.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to