You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by "ASF GitHub Bot (Jira)" <ji...@apache.org> on 2026/10/18 20:56:00 UTC
[jira] [Updated] (HDDS-17113) SCM to prefer a Ratis pipeline on the writer's datanode when allocating blocks
[ https://issues.apache.org/jira/browse/HDDS-17113?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
ASF GitHub Bot updated HDDS-17113:
----------------------------------
Labels: pull-request-available (was: )
> SCM to prefer a Ratis pipeline on the writer's datanode when allocating blocks
> ------------------------------------------------------------------------------
>
> Key: HDDS-17113
> URL: https://issues.apache.org/jira/browse/HDDS-17113
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Siyao Meng
> Assignee: Siyao Meng
> Priority: Major
> Labels: pull-request-available
>
> When a client runs on a datanode host (for example an HBase RegionServer colocated with a datanode), Ratis block allocation ignores the client's location, so none of the pipeline's datanodes may be local to the writer. HDFS places the first replica on the writer's datanode, which lets later reads of that data (e.g. HBase compactions) use short-circuit reads.
> Proposal: add {{hdds.scm.block.allocation.prefer.client.local}} (default {{false}}). When set, OM passes the client address to SCM on every block allocation, and SCM first tries open Ratis pipelines led by the client's datanode, then other open pipelines that contain it, before falling back to the pipeline choose policy. A pipeline led by the client's datanode also keeps the client to leader hop of every write on the same host, which is the colocation discussed in HDDS-11000.
> EC allocation is unchanged.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org