[ https://issues.apache.org/jira/browse/HDFS-9173?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14951004#comment-14951004 ]
Jing Zhao commented on HDFS-9173: --------------------------------- Thanks a lot for working on this, [~walter.k.su]! As [~zhz] suggested, to use "the smallest length that covers at least 6 internal blocks" as the safe length can be a good start point. In some other scenarios, we can save more data though. E.g., if the data looks like: {code} blk_0 blk_1 blk_2 blk_3 blk_4 blk_5 blk_6 blk_7 blk_8 64k 64k 64k 64k 64k 64k 64k 64k 64k 64k 13k ___ ___ ___ ___ 64k 64k 64k {code} i.e., the failure happens when the client tries to close the file. But I'm fine if we do not handle this kind of scenario right now. For the patch, do you think we can divide it into several smaller ones? Currently I'm thinking maybe we can separating the safe length calculation out into another jira, and add more tests there. These tests can only focus on the calculation thus do not need to be end-to-end, and in this way we can cover more scenarios. I will review the whole patch and post comments later. > Erasure Coding: Lease recovery for striped file > ----------------------------------------------- > > Key: HDFS-9173 > URL: https://issues.apache.org/jira/browse/HDFS-9173 > Project: Hadoop HDFS > Issue Type: Sub-task > Reporter: Walter Su > Assignee: Walter Su > Attachments: HDFS-9173.00.wip.patch, HDFS-9173.01.patch > > -- This message was sent by Atlassian JIRA (v6.3.4#6332)