[
https://issues.apache.org/jira/browse/HDFS-15981?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
zhanghuazong updated HDFS-15981:
--------------------------------
Description:
When the block report satisfies the block distribution strategy, the block is
removed from the lowredundancyBlocks. But removing the block from the
lowredundancyBlocks is a redundant operation.
First, in the patch queue, the block removal operation will be performed in the
method chooseSourceDatanodes and validateReconstructionWork.
second, the removal of the block report will only be at the
QUEUE_REPLICAS_BADLY_DISTRIBUTED level, which is not an accurate operation.
Finally, when there is a large amount of data in the
QUEUE_REPLICAS_BADLY_DISTRIBUTED queue, the processing efficiency of the block
report will be reduced
was:
When the block report satisfies the block distribution strategy, the block is
removed from the lowredundancyBlocks. But removing the block from the
lowredundancyBlocks is a redundant operation.
First, in the patch queue, the block removal operation will be performed in the
method chooseSourceDatanodes and validateReconstructionWork.
second, the removal of the block report will only be at the
QUEUE_REPLICAS_BADLY_DISTRIBUTED level, which is not an accurate operation.
> Removing redundant block queues will slow down block reporting
> --------------------------------------------------------------
>
> Key: HDFS-15981
> URL: https://issues.apache.org/jira/browse/HDFS-15981
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs
> Reporter: zhanghuazong
> Assignee: zhanghuazong
> Priority: Major
>
> When the block report satisfies the block distribution strategy, the block is
> removed from the lowredundancyBlocks. But removing the block from the
> lowredundancyBlocks is a redundant operation.
> First, in the patch queue, the block removal operation will be performed in
> the method chooseSourceDatanodes and validateReconstructionWork.
> second, the removal of the block report will only be at the
> QUEUE_REPLICAS_BADLY_DISTRIBUTED level, which is not an accurate operation.
> Finally, when there is a large amount of data in the
> QUEUE_REPLICAS_BADLY_DISTRIBUTED queue, the processing efficiency of the
> block report will be reduced
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
