[ https://issues.apache.org/jira/browse/HDFS-5633?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13841180#comment-13841180 ]
Hudson commented on HDFS-5633: ------------------------------ FAILURE: Integrated in Hadoop-Yarn-trunk #413 (See [https://builds.apache.org/job/Hadoop-Yarn-trunk/413/]) HDFS-5633. Improve OfflineImageViewer to use less memory. Contributed by Jing Zhao. (jing9: http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1548359) * /hadoop/common/trunk/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt * /hadoop/common/trunk/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/tools/offlineImageViewer/FileDistributionVisitor.java * /hadoop/common/trunk/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/tools/offlineImageViewer/ImageLoaderCurrent.java > Improve OfflineImageViewer to use less memory > --------------------------------------------- > > Key: HDFS-5633 > URL: https://issues.apache.org/jira/browse/HDFS-5633 > Project: Hadoop HDFS > Issue Type: Improvement > Reporter: Jing Zhao > Assignee: Jing Zhao > Priority: Minor > Fix For: 2.4.0 > > Attachments: HDFS-5633.000.patch > > > Currently after we rename a file/dir which is included in a snapshot, the > file/dir can be linked with two different reference INodes. To avoid > saving/loading the inode multiple times in/from FSImage, we use a temporary > map to record whether we have visited this inode before. > However, in OfflineImageViewer (specifically, in ImageLoaderCurrent), the > current implementation simply records all the directory inodes. This can take > a lot of memory when the fsimage is big. We should only record an inode in > the temp map when it is referenced by an INodeReference, just like what we do > in FSImageFormat. -- This message was sent by Atlassian JIRA (v6.1#6144)