[
https://issues.apache.org/jira/browse/HDDS-4061?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated HDDS-4061:
---------------------------------
Labels: pull-request-available (was: )
> Pending delete blocks are not always included in #BLOCKCOUNT metadata
> ---------------------------------------------------------------------
>
> Key: HDDS-4061
> URL: https://issues.apache.org/jira/browse/HDDS-4061
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: Ozone Datanode
> Reporter: Ethan Rose
> Assignee: Ethan Rose
> Priority: Minor
> Labels: pull-request-available
>
> Within the code that handles containers, there is currently some disagreement
> as to whether blocks marked for deletion with the #deleting# prefix should be
> included in the #BLOCKCOUNT metadata value. The BlockDeletingService includes
> pending delete blocks in the block count, since after deleting the blocks,
> BlockDeletingService#call calls
> KeyValueContainerData#updateAndCommitDBCounters, where the pending deletion
> blocks (now deleted) are subtracted from the block count metadata. However,
> KeyValueContainerUtil#initializeUsedBytesAndBlockCount filters away all
> blocks that have prefixes when setting the block count. This means pending
> delete blocks with the #deleting# prefix are excluded from the block count.
> This fix will alter KeyValueContainerUtil#initializeUsedBytesAndBlockCount to
> include #deleting# blocks when setting the #BLOCKCOUNT and #BYTESUSED
> metadata values. It will also adjust the TestContainerReader unit tests to
> account for this change.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
