[
https://issues.apache.org/jira/browse/HDDS-11777?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ethan Rose updated HDDS-11777:
------------------------------
Summary: Create common method for tasks that need to iterate OM namespace
in parallel (was: Common code for reverse mapping db)
> Create common method for tasks that need to iterate OM namespace in parallel
> ----------------------------------------------------------------------------
>
> Key: HDDS-11777
> URL: https://issues.apache.org/jira/browse/HDDS-11777
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Sumit Agrawal
> Priority: Major
>
> Add a utility method that supports iterating the filesystem tree from RocksDB
> in a parallel manner. This has multiple potential use cases, including
> building a container to key mapping from a debug tool, repairing quota
> issues, and namespace processing on Recon snapshot install, like DU and
> container to key mapping.
> This method would be used in cases where we can sacrifice in-order traversal
> of keys to gain parallelization with low memory overhead, using a modified
> DFS with parallel tree splitting for FSO and more simple work splitting for
> OBS. Signature would look something like this:
> {code:java}
> void walkNamespace(DB omDB, String startPath, Consumer<KeyInfo> action, int
> threadCount)
> {code}
> Each use case would specify the action it needs to perform on each key in the
> consumer.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
