[
https://issues.apache.org/jira/browse/HDDS-15074?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wei-Chiu Chuang updated HDDS-15074:
-----------------------------------
Resolution: Won't Do
Status: Resolved (was: Patch Available)
> [DataNode] Implement lookahead dispatcher in ReplicationSupervisor
> ------------------------------------------------------------------
>
> Key: HDDS-15074
> URL: https://issues.apache.org/jira/browse/HDDS-15074
> Project: Apache Ozone
> Issue Type: Task
> Reporter: Wei-Chiu Chuang
> Assignee: Wei-Chiu Chuang
> Priority: Major
> Labels: pull-request-available
>
> Refactor the ReplicationSupervisor to use a volume-aware dispatching logic.
> Instead of a simple priority pull, the dispatcher should "look ahead" in the
> priority queue to skip tasks targeting busy volumes (those at the
> volume.outbound.limit) while keeping them at the head of the queue. Implement
> a wake-up
> mechanism to re-evaluate the queue whenever a volume slot becomes
> available, ensuring high node throughput without starvation or disk thrashing.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
