[
https://issues.apache.org/jira/browse/HDFS-11545?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15931954#comment-15931954
]
Arpit Agarwal commented on HDFS-11545:
--------------------------------------
Thanks [~hanishakoneru]. The patch looks great. Minor feedback:
# The following comment in DatanodeProtocol.proto looks unnecessary since
SlowDiskReportProto does not include the DataNodeId.
{code}
* DataNode. (DataNodeId must be transmitted as a string for protocol
* compability with earlier versions of Hadoop).
{code}
# NUM_DISK_OPS can be initialized to {{DiskOp.values().length()}} instead of
hard-coding 3.
# Extra space in {{return slowDiskInfoProtos}}.
We should add a few unit tests:
# Test for PBHelper serialization/deserialization. See
TestPBHelper#testSlowPeerInfoPBHelper.
# Ensure that the datanode sends out a slow disk report when one is due, and it
includes the contents of {{dn.getDiskMetrics().getDiskOutliersStats()}}.
> Propagate DataNode's slow disks info to the NameNode via Heartbeat
> ------------------------------------------------------------------
>
> Key: HDFS-11545
> URL: https://issues.apache.org/jira/browse/HDFS-11545
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Hanisha Koneru
> Assignee: Hanisha Koneru
> Attachments: HDFS-11545.001.patch
>
>
> DataNode detects the outliers (slow disks) among all its disk. This
> information can be propagated to the NameNode so that the NameNode gets disk
> information from all Datanodes.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
