Skip to content

Instantly share code, notes, and snippets.

@dvryaboy
Created April 28, 2011 01:03

Revisions

  1. dvryaboy created this gist Apr 28, 2011.
    19 changes: 19 additions & 0 deletions kill_hadoop_orphans.sh
    Original file line number Diff line number Diff line change
    @@ -0,0 +1,19 @@
    #!/bin/bash

    # Kill tasks holding on to deleted userlogs. These are most likely abandoned jobs.

    function get_bad_pids {
    for i in `ps -ef | grep java | awk '{print $2;}'`; do
    cnt=`/usr/sbin/lsof -p $i | grep deleted | grep /var/log/hadoop-0.20/userlogs/attempt | wc -l`;
    if [ $cnt -gt 0 ]; then
    PIDS=$i:$PIDS;
    fi
    done
    }

    PIDS=""
    get_bad_pids
    IFS=':'
    for pid in $PIDS; do
    kill -9 $pid;
    done