[ https://issues.apache.org/jira/browse/HADOOP-11566?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14532487#comment-14532487 ]
Kai Zheng commented on HADOOP-11566: ------------------------------------ Hi Vinay, Thanks for your good question. Yes the order is important as the underlying raw decoder requires that. The order is: parity units first, then data units, as consistent with raw decoder. Please note the input units and output units are guided by the {{erasedIndexes}} which is carefully computed as indexes into the inputs array (parity units + data units). > Add tests and fix for erasure coders to recover erased parity units > -------------------------------------------------------------------- > > Key: HADOOP-11566 > URL: https://issues.apache.org/jira/browse/HADOOP-11566 > Project: Hadoop Common > Issue Type: Sub-task > Reporter: Kai Zheng > Assignee: Kai Zheng > Attachments: HADOOP-11566-HDFS-7285-v2.patch, HADOOP-11566-v1.patch > > > Discussing with [~zhz] in HADOOP-11542: it's planned to have follow up a JIRA > to enhance the tests for parity chunks as well. Like erasedDataIndexes, > erasedParityIndexes will be added to specify which parity units are to be > erased and recovered then. -- This message was sent by Atlassian JIRA (v6.3.4#6332)