[ https://issues.apache.org/jira/browse/HDFS-7740?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14329683#comment-14329683 ]
Konstantin Shvachko commented on HDFS-7740: ------------------------------------------- This actually worked pretty good. Only one test is running too long: {{testTruncateWithDataNodesShutdownImmediately()}} adds 30 secs to the running time because you check {{isUnderConstruction()}} for the block 300 times with interval 100 msec. I'd suggest waiting for DNs being down, and then checking the block being still under construction. {code} cluster.shutdownDataNodes(); try { for(int i = 0; i < SUCCESS_ATTEMPTS && cluster.isDataNodeUp(); i++) { Thread.sleep(SLEEP); } assertFalse("All DataNodes should be down.", cluster.isDataNodeUp()); LocatedBlocks blocks = getLocatedBlocks(p); assertTrue(blocks.isUnderConstruction()); } finally { cluster.startDataNodes(conf, DATANODE_NUM, true, StartupOption.REGULAR, null); cluster.waitActive(); } {code} I am +1 on the rest. Yi, if you cannot update the patch I can make just this change to the latest and commit. > Test truncate with DataNodes restarting > --------------------------------------- > > Key: HDFS-7740 > URL: https://issues.apache.org/jira/browse/HDFS-7740 > Project: Hadoop HDFS > Issue Type: Sub-task > Components: test > Affects Versions: 2.7.0 > Reporter: Konstantin Shvachko > Assignee: Yi Liu > Fix For: 2.7.0 > > Attachments: HDFS-7740.001.patch, HDFS-7740.002.patch > > > Add a test case, which ensures replica consistency when DNs are failing and > restarting. -- This message was sent by Atlassian JIRA (v6.3.4#6332)