sodonnel commented on PR #4655:
URL: https://github.com/apache/ozone/pull/4655#issuecomment-1589666709

   Where are we with getting this committed? Its been open a long time now.
   
   Also, I am concerned the checks on the DN to remove a container are too 
conservative. If RocksDB has no entry for a block, I am not sure we should by 
default check for chunks on disk - with no RocksDB data we cannot access them 
anyway and the checksum data will be lost (it is stored in RocksDB).
   
   See my comment earlier today and also more detail in 
https://issues.apache.org/jira/browse/HDDS-8838. This is an example of a 
container that should be deleted, but it will not delete (even before this 
change, but I don't think this change will alter that behavior).
   
   I also wonder if the default state for a containerData object should be 
"empty" and then it gets toggled to "non empty" on the first successful put 
block to it - that may be one way to catch this scenario, but also it is 
possible for a container to have an orphan block like this, and other blocks. 
The other blocks could get removed over time, leaving the container empty 
except for this. orphan block.


-- 
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