copy

# -*- mode: sh;-*-

# Copy all gpfs2 home and shared data to /scratch/gpfs2_full_backup.

# Global variables
#-----------------
file_dirlist=./copy_input_dirs
file_sentinel=ORIG_COPY_TO_SCRATCH_COMPLETED
dir_logs=logs
log=copy.log
joblog=copy.joblog
np=6				# parallel rsync processes to run.
prefix_src=/gpfs/gpfs2
prefix_dest=/gpfs/scratchfs1/gpfs2_full_backup
parallel=apps/bin/parallel
simulate=			# test without copying.

# Functions
#----------

# Generate file with list of directories to be copied.  This is
# necessary to "parallelize" rsync.
create_dirlist() {
    # Find does not order directories, so do not generate the file if
    # it already exists to preserve the copying order and support
    # predictable resuming.
    if ! [[ -f $file_list ]]; then
	find $prefix_src/{home,shared} \
	    -mindepth 1 \
	    -maxdepth 1 \
	    -type d \
	    > $file_dirlist
    fi
    # Ensure target sub-directories exist with same ownership,
    # permissions and modification times as source sub-directories.
    for dir in home shared; do
	src=$prefix_src/$dir
	dest=$prefix_dest/$dir
        mkdir -p $dest
	chown --reference=$src $dest
	chmod --reference=$src $dest
	touch --reference=$src $dest
    done
}

# Prefix timestamp before "echo" message
msg() {
    echo "[$(date '+%Y-%m-%dT%H:%M:%S')] $1"
}

# Copy a single directory.
copy_dir() {
    local src dest
    src=$1
    dest=${src/#$prefix_src/$prefix_dest}
    
    msg "Started copying directory $src -> $dest"
    if ! [[ -z "$simulate" ]]; then
	sleep $(( ( $RANDOM % 3 )  + 1 ))s
    else
	path_sentinel=$dest/$file_sentinel
	if [[ -f $path_sentinel ]]; then
	    msg "Skipping already copied directory $dest"
        else
	    rsync -aX $src $dest 2>&1
	    touch $path_sentinel
	fi
    fi
    msg "Finished copying directory $src -> $dest"
}

# Use GNU parallel to parallelize copying instead of xargs, because
# GNU parallel has nicer support for running exported bash functions
# which allows us to have a single, self-contained script.
copy_dirs() {
    # Export bash functions and variables for GNU parallel to inherit
    # with --env.
    export -f copy_dir msg
    export simulate prefix_src prefix_dest file_sentinel
    # Use --line-buffer so that output lines don't split.
    cat $file_dirlist |
    $parallel \
	--will-cite \
	--env \
	--line-buffer \
	--max-args 1 \
	--max-procs $np \
	--output-as-files \
	--results $dir_logs/{} \
	--eta \
	--joblog $joblog \
	copy_dir
}

# Main
#-----

main() {
    msg "---" 
    msg "BEGIN copy"
    create_dirlist
    copy_dirs
    msg "END copy"
}

# Boilerplate for running `main()`.
[[ "$0" != "$BASH_SOURCE" ]] || main "$@" #|& tee -a $log