[ https://issues.apache.org/jira/browse/HDFS-3533?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13396191#comment-13396191 ]
Todd Lipcon commented on HDFS-3533: ----------------------------------- I think this still persists in 1.x -- it looks like it could happen with the following race: - client gets block locations for an over-replicated block, where the NN has enqueued a command to one DN to delete its extra replica - the command reaches the DN and it deletes the block - the client issues primary.getBlockInfo(...) but the block doesn't exist, so it returns null - client gets NPE This might also happen if the block's genstamp is changing (perhaps when reading a file under construction where the pipeline has just been recovered?) > NPE in DFSClient$DFSInputStream.openInfo > ---------------------------------------- > > Key: HDFS-3533 > URL: https://issues.apache.org/jira/browse/HDFS-3533 > Project: Hadoop HDFS > Issue Type: Bug > Components: hdfs client > Affects Versions: 0.20.203.0 > Reporter: Amareshwari Sriramadasu > > I'm seeing the following NPE: > java.lang.NullPointerException > at > org.apache.hadoop.hdfs.DFSClient$DFSInputStream.openInfo(DFSClient.java:1633) > at > org.apache.hadoop.hdfs.DFSClient$DFSInputStream.<init>(DFSClient.java:1593) > at org.apache.hadoop.hdfs.DFSClient.open(DFSClient.java:428) > at > org.apache.hadoop.hdfs.DistributedFileSystem.open(DistributedFileSystem.java:187) > at org.apache.hadoop.fs.FileSystem.open(FileSystem.java:456) > Here, the file gets opened while it's being written. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira