Basic functions to do hadoop (hdfs) health check #bigdata #hadoop #bash #hdfs

hadoop_health_check.sh

#!/bin/bash 

function emsg() { email_body="$email_body\n${1}"; }


function email()
{
        local subject=$1 rcpt=$2
        echo -e "$email_body\n\n\n" | mail -s "$subject" -r [email protected] "$rcpt"
}

function hadoop_nn_state()
{
        local h=$1 ret state
        state=`curl -s "http://$h:50070/jmx?qry=Hadoop:service=NameNode,name=NameNodeStatus" | grep '\"State\" :'`
        ret=$?
        [ $ret -ne 0 ] &&  { emsg "Hadoop $h name node problem, curl returned $ret"; return $ret; }
        state=${state//\"/}
        state=${state//,/}
        state=${state##*:}
        state=${state// /}

        [ x$state != x'active'  ] &&  { emsg "Hadoop $h name node problem, it is not active, current state: $state"; return 1; }
        dbg "Hadoop NameNode: $h state is $state"
        return 0
}

function cluster_status()
{
        local a nr dnodes=$1
        while read a; do
                echo $a | grep -q "Live datanodes"
                [ $? -ne 0 ] && continue
                nr=${a##*(}
                nr=${nr%%)*}
                [ $nr -ne $dnodes ] && { emsg "Hadoop cluster problem: live datanodes $nr ,$a"; return ; }

        done < <(HADOOP_USER_NAME=hdfs /usr/local/hadoop/bin/hadoop dfsadmin -report 2>/dev/null)
}

hadoop_nn_state 192.168.1.1
cluster_status 3 # arg is number of data nodes

[ "$email_body" != '' ] && email 'HDFS problem' [email protected]