[ https://issues.apache.org/jira/browse/HDFS-8383?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Walter Su updated HDFS-8383: ---------------------------- Attachment: HDFS-8383.01.patch Thanks Zhe for your very good advices. I moved {{BlockRecoveryTrigger}} inside {{Coordinator}} as you said. I have a feel {{Coordinator}} can do other jobs, So I make {{Coordinator}} a service. I add the javadocs your posted and changed a little bit. Hope you are ok with that. bq. Why not incrementing numScheduled if it's already positive? I make it a {{boolean}}. The thing is, assume horizontal axis represents time, {noformat} streamer#1 failed --> scheduled ----->recovery happening (would bump GS to 1002) streamer#2 failed --> scheduled ----> waiting streamer#3 failed --> scheduled --> waiting {noformat} the failures of #2,#3 can be processed together and we bump GS to 1003. We only need schedule once. > Tolerate multiple failures in DFSStripedOutputStream > ---------------------------------------------------- > > Key: HDFS-8383 > URL: https://issues.apache.org/jira/browse/HDFS-8383 > Project: Hadoop HDFS > Issue Type: Sub-task > Reporter: Tsz Wo Nicholas Sze > Assignee: Walter Su > Attachments: HDFS-8383.00.patch, HDFS-8383.01.patch > > -- This message was sent by Atlassian JIRA (v6.3.4#6332)