duongkame commented on PR #5694:
URL: https://github.com/apache/ozone/pull/5694#issuecomment-1857354371

   > > @sumitagrawl I think you are also describing what I wrote on the Jira 
[here](https://issues.apache.org/jira/browse/HDDS-9388?focusedCommentId=17776368&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-17776368).
 IMO this is more of a minor change in behavior than an issue. The server is 
still failing the request based on an ordering either way.
   > 
   > @errose28 yes, 2 cases we are changing,
   > 
   > 1. ACL as already mentioned
   > 2. linked bucket resolution also
   
   @sumitagrawl 
   I don't see problems with Link resolution in preExecute. E.g. at the time a 
key-create request reaches the leader node, the request should be rejected if 
the link is not there. This is different from before when the link creation can 
happen after a while before the key-creation request reaches each replica. A 
reverse example is when a link is being removed. Either case, making the 
decision at the leader makes sense to me.
   
   Yes, there are changes in behaviors the edge cases but they're not 
necessarily issues. 
   
   The same story for moving ACL checks, as @errose28 mentioned.
   
   I think moving validation/pre-processing out of applyTransaction is the 
right direction. ApplyTransaction should be to apply the state of a request to 
the state machine. 
   
   C.c. some more folks as I need your opinions too. @kerneltime @szetszwo 
@smengcl.
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to