rebalance_large_files

rebalance_large_files.sh

#!/bin/bash

list_large_files_checksum ()
{
    local DIR=$1;
    large_files=""
    large_directories=""
    # Get top ten culprit files
    for i in $(du --exclude=".glusterfs" -akx $DIR | sort -nr | head -10 | awk {'print $2'}); do
        if [ -f $i ]; then
            large_files="$large_files $i"
        else
            large_directories="$large_directories $i"
        fi
    done

    if [ -z "$large_files" ]; then
        echo "No large files found, but there are large directories still"
        echo "$large_directories"
        exit 1;
    fi
    echo -n "Generating checksum for $large_files"
    for i in $large_files; do
        if [ -f $i ]; then
            md5sum $i | awk {'print $1'} > /tmp/$(basename $i).md5sum
        fi;
    done
    echo " Done."
}

mount_glusterfs ()
{
    local volname="$1"
    local mount_point="$2";

    if [ -f "/var/lib/glusterd/vols/$volname/$volname-fuse.vol" ]; then
        subvolumes_replicated=$(grep -w 'volume' /var/lib/glusterd/vols/$volname/$volname-fuse.vol | grep replicate | awk {\
'print $2'})
    fi

    while true; do
        echo "Replicated subvolumes found are $subvolumes_replicated";
        echo -n "Enter your subvolume name to mount and press [ENTER]: "
        read subvolumename

        echo -n "Mouting $subvolumename from $volname..";
        mount -t glusterfs -ovolume-name=$subvolumename localhost:/$volname $mount_point;
        echo "Done."
        sleep 2;

        echo -n "Enter absolute path of the filename to be transferred and press [ENTER]: "
        read filename

        relative_file=$(echo $filename | sed 's/PATTERN//g');
        relative_dir=$(dirname $relative_file)

        echo -n "Copying $filename to $relative_dir inside $mount_point ..";
        rsync -at $filename $mount_point/$relative_dir;
        echo "Done."

        echo "Verifying.. from old checksum"
        old_md5sum=$(cat /tmp/$(basename $filename).md5sum)
        new_md5sum=$(md5sum $mount_point/$relative_file |awk {'print $1'});

        umount $mount_point;
        if [ $old_md5sum != $new_md5sum ]; then
            echo "Catastrophic error exiting corruption found";
            exit 1;
        fi
        echo -n "Are you done transferring? [y/n]:"
        read yes
        if [ $yes == "y" ]; then
            exit 1;
        fi
    done
}

main ()
{
    if [ $# -ne 3 ]; then
        echo "wrong number of arguments given"
        echo " $0 <large_file_checksum_dir> <volumename> <mount_point>"
        exit 1;
    fi

    list_large_files_checksum $1;
    mount_glusterfs $2 $3;
}

main "$@";