[ https://issues.apache.org/jira/browse/HDFS-8873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14876099#comment-14876099 ]
Colin Patrick McCabe commented on HDFS-8873: -------------------------------------------- bq. Question about the rest of your comments: Except for the last one, the rest of your comments apply to code that I touched only by indenting it. I didn't modify any of that code that I didn't have to, in the name of simplifying review by diff. Sounds like the going rule is, "you touch it, you own it," even if it's just a whitespace change. Am I reading that correctly? That's a fair point. If you want to leave the d and m variables as-is to simplify the diff and backporting, that's fine. We should fix them at some point, but it doesn't have to be right now. > throttle directoryScanner > ------------------------- > > Key: HDFS-8873 > URL: https://issues.apache.org/jira/browse/HDFS-8873 > Project: Hadoop HDFS > Issue Type: Improvement > Components: datanode > Affects Versions: 2.7.1 > Reporter: Nathan Roberts > Assignee: Daniel Templeton > Attachments: HDFS-8873.001.patch, HDFS-8873.002.patch, > HDFS-8873.003.patch > > > The new 2-level directory layout can make directory scans expensive in terms > of disk seeks (see HDFS-8791) for details. > It would be good if the directoryScanner() had a configurable duty cycle that > would reduce its impact on disk performance (much like the approach in > HDFS-8617). > Without such a throttle, disks can go 100% busy for many minutes at a time > (assuming the common case of all inodes in cache but no directory blocks > cached, 64K seeks are required for full directory listing which translates to > 655 seconds) -- This message was sent by Atlassian JIRA (v6.3.4#6332)