[ https://issues.apache.org/jira/browse/HDFS-8025?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ming Ma updated HDFS-8025: -------------------------- Attachment: HDFS-8025.patch The reason unit test "TestDecommission#testDecommissionWithNamenodeRestart" can pass because blockReport was processed earlier than DecommissonManager thread. Thus the test case doesn't verify the scenario. The patch added {{checkBlockReportReceived}} back. It also improves unit tests in the following ways. a) increase the blockreport interval and make it more likely for DecommissonManager thread to process the node before blockReport is processed. b) Check if the blocks are properly replicated right after node is transitioned to Decommissioned. There is no need to wait for the block replication given nodes should only be transitioned to Decommissioned after block are properly replicated. > Addendum for HDFS-3087 > ---------------------- > > Key: HDFS-8025 > URL: https://issues.apache.org/jira/browse/HDFS-8025 > Project: Hadoop HDFS > Issue Type: Bug > Reporter: Ming Ma > Attachments: HDFS-8025.patch > > > Per discussion with [~andrew.wang] on HDFS-7411, we should include HDFS-3087 > and enhance the unit test. -- This message was sent by Atlassian JIRA (v6.3.4#6332)