[ https://issues.apache.org/jira/browse/HDFS-6982?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14118971#comment-14118971 ]
Maysam Yabandeh commented on HDFS-6982: --------------------------------------- Thanks [~philip]. I agree with you. I actually was planning to skip the audit log tailing stuff altogether to keep the patch simple. If there was interest in future I can submit a separate patch for that. The metric key format is "operation.user". Here is a sample output from the jmx interface: {code} [myabandeh@smf1-aro-39-sr1(hadoop-tst-nn) ~]$ curl localhost:12333/jmx | grep Hadoop:service=nntop,name=topusers -B1 -A8 % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 }, { "name" : "Hadoop:service=nntop,name=topusers", "modelerType" : "topusers", "tag.Context" : "namenode", "tag.ProcessName" : "DummyProcessName", "tag.SessionId" : "DummySessionId", "tag.Hostname" : "hhh", "delete.xxx" : 1, "setPermission.ALL" : 0, "getfileinfo.ALL" : 3159, {code} > nntop: top-like tool for name node users > ----------------------------------------- > > Key: HDFS-6982 > URL: https://issues.apache.org/jira/browse/HDFS-6982 > Project: Hadoop HDFS > Issue Type: New Feature > Reporter: Maysam Yabandeh > Attachments: nntop-design-v1.pdf > > > In this jira we motivate the need for nntop, a tool that, similarly to what > top does in Linux, gives the list of top users of the HDFS name node and > gives insight about which users are sending majority of each traffic type to > the name node. This information turns out to be the most critical when the > name node is under pressure and the HDFS admin needs to know which user is > hammering the name node and with what kind of requests. Here we present the > design of nntop which has been in production at Twitter in the past 10 > months. nntop proved to have low cpu overhead (< 2% in a cluster of 4K > nodes), low memory footprint (less than a few MB), and quite efficient for > the write path (only two hash lookup for updating a metric). -- This message was sent by Atlassian JIRA (v6.3.4#6332)