[ https://issues.apache.org/jira/browse/HDFS-10917?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15771091#comment-15771091 ]
Xiaoyu Yao commented on HDFS-10917: ----------------------------------- Thanks [~xiaobingo] for working on this. Patch v04 looks good to me. I just have a few questions: BlockReceiver.java 1. Line 828: can you clarify the design behind tracking penultimate node only in a pipeline line [dn0, dn1, dn2]? It works well for example like: 5, 100, 105. What if we have latency of pipeline like: 5, 115, 15? Do we miss detecting non-penultimate slow datanodes? DataNodeMXBean.java 2. Line 130: we are getting the average info of SendPacketDownstream when acting as penultimate (2nd to last) node. RollingAverages.java 3. NIT: Line 63: should be .setNameFormat("RollingAverages-%d") > Collect peer performance statistics on DataNode. > ------------------------------------------------ > > Key: HDFS-10917 > URL: https://issues.apache.org/jira/browse/HDFS-10917 > Project: Hadoop HDFS > Issue Type: Improvement > Components: datanode > Affects Versions: 2.8.0 > Reporter: Xiaobing Zhou > Assignee: Xiaobing Zhou > Attachments: HDFS-10917.000.patch, HDFS-10917.001.patch, > HDFS-10917.002.patch, HDFS-10917.003.patch, HDFS-10917.004.patch > > > DataNodes already detect if replication pipeline operations are slow and log > warnings. For the purpose of analysis, performance metrics are desirable. > This proposes adding them on DataNodes. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org