Check HDFS for a specific file a certain amount of times before it errors out so you don't execute code that has a dependency.

gistfile1.txt

CHECK_HDFS="/some/path/to/file"

function hdfsCheck {
    RETRY=0

    while [ $RETRY -lt 9 ];
    do
        COUNT=$(hdfs dfs -ls "${CHECK_HDFS}" | wc -l) 2> stderr.txt

        if [ $COUNT -lt 1 ]; then
            echo "Try $RETRY. Replication hasn't happened yet! $(date +"%Y-%m-%d %H:%M")"
            sleep 20m
            RETRY=$[ $RETRY + 1 ]
        else
            echo "Replication occured."
            break
        fi
    done

    if [ $RETRY -lt 9 ]; then
        return 0
    else
        return 1
    fi
}