[ 
https://issues.apache.org/jira/browse/SENTRY-2126?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

kalyan kumar kalvagadda updated SENTRY-2126:
--------------------------------------------
    Description: 
Transactional reties not needed for failures while persisting 
MSentryHmsNotification which are because of duplicate key. 

 

 

There are only inserts that happen on this table. If an insert failed because 
of duplicate entry, it will likely fill in the subsequent retires. There is no 
need to retry.

 

Specially in HA mode when multiple sentry servers can act as leader (split 
Brian situation) this will cause additional load on the database for no reason.

  was:
Transactional reties not needed for failures while persisting 
MSentryHmsNotification which are because of duplicate key. 

 

 

There are only inserts that happen on this table. If an insert failed because 
of duplicate entry, it will likely fill in the subsequent retires. There is not 
no need to retry.

 

Specially in HA mode when multiple sentry servers can act as leader (split 
Brian situation) this will cause additional load on the database for no reason.


> Transactional retry not needed for JDO exceptions on 
> SENTRY_HMS_NOTIFICATION_ID
> -------------------------------------------------------------------------------
>
>                 Key: SENTRY-2126
>                 URL: https://issues.apache.org/jira/browse/SENTRY-2126
>             Project: Sentry
>          Issue Type: Bug
>          Components: Sentry
>    Affects Versions: 2.1.0
>            Reporter: kalyan kumar kalvagadda
>            Assignee: kalyan kumar kalvagadda
>            Priority: Major
>
> Transactional reties not needed for failures while persisting 
> MSentryHmsNotification which are because of duplicate key. 
>  
>  
> There are only inserts that happen on this table. If an insert failed because 
> of duplicate entry, it will likely fill in the subsequent retires. There is 
> no need to retry.
>  
> Specially in HA mode when multiple sentry servers can act as leader (split 
> Brian situation) this will cause additional load on the database for no 
> reason.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to