[
https://issues.apache.org/jira/browse/HDDS-12628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17936621#comment-17936621
]
Stephen O'Donnell commented on HDDS-12628:
------------------------------------------
This is already implemented. The under replicated queue is defined as:
{code}
underRepQueue = Queues.synchronizedQueue(new PriorityQueue<>(
Comparator.comparing(ContainerHealthResult
.UnderReplicatedHealthResult::getWeightedRedundancy)
.thenComparing(ContainerHealthResult
.UnderReplicatedHealthResult::getRequeueCount)));
{code}
And the weighted redundancy will start at 2 if there are two replicates
remaining, or 1 if there is only 1. Under replication caused by decommissioning
replicas are given a lower priority and will only be attempted after the
"reduced redundancy" replicas are dealt with.
> Prioritize the most under replicated containers for re-replication
> ------------------------------------------------------------------
>
> Key: HDDS-12628
> URL: https://issues.apache.org/jira/browse/HDDS-12628
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Wei-Chiu Chuang
> Priority: Major
>
> HDDS-6957 implemented the logic to prioritize under replicated containers.
> However it does not further prioritize those that are minimally replicated
> (excluding: decommissioned nodes and corrupt replicas) and thus in danger of
> missing permanently.
> As a matter of fact, HDFS implements that logic in LowRedundancyBlocks
> [https://github.com/apache/hadoop/blob/trunk/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/LowRedundancyBlocks.java]
> The same logic could be applicable to Ozone container replication.
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
