An rclone backup script for cron

rclone-cron.sh

#!/bin/bash

##############################################################################
# An rclone backup script by Chris Tippett ([email protected])
#
# Originally adapted from the great work by Jared Males ([email protected])
# https://gist.github.com/jaredmales/2f732254bb10002fc0549fa9aa0abdd7
# 
# Copyright (C) 2020 Chris Tippett ([email protected])
#
# This script is licensed under the terms of the MIT license.
# https://opensource.org/licenses/MIT
#
# Runs the 'rclone sync' command. Designed to be used as a cron job.
#
# 1) Backup source
#    Edit the $src variable below to point to a local directory or remote location.
#
# 2) Backup destination
#    Edit the $dest variable to point to a local or remote (see rclone docs).
# 
# 3) Excluding files and directories
#    Edit the $opt_exclude_file variable below to point to a file listing files and directories to exclude.
#    See: https://rclone.org/filtering/
#
#    Also, any directory can be excluded by adding an '.rclone-ignore' file to it without editing the exclude file.
#    This file can be empty.  You can edit the location of this file with the RCLONE_EXCLUDE_FILE environment variable.
# 
# 4) You can change the bandwidth limits by editing $opt_bwlimit, which includes a timetable facility.  
#    See: https://rclone.org/docs/#bwlimit-bandwidth-spec
#
# 5) Logs:
#    -- By default rclone will log its output to the '.rclone' sub-directory under either $src or $dest (depending
#       on the one that is local to your filesystem).
#    -- The log filename is `rclone-<remotename>.log`, this is rotated using savelog.
#    -- The output of this script (run by cron) is written to stdout. This can be redirected to a location of your
#       choosing from within crontab.
#
# 6) Example crontab
#     */1 * * * * /home/johndoe/.config/rclone/rclone-cron.sh sync "/home/johndoe" "gdrive:" >> /var/log/rclone-cron.log 2>&1
#   
##############################################################################

### CONFIGURATION
#
# input arguments
src="${1}" # source
dest="${2}" # destination

# optional
# log_path="/var/log/" # override default log with your own location

# other options - https://rclone.org/flags/
RCLONE_EXCLUDE_FILE="$(dirname "$0")/exclude-file.txt" # read file exclusion patterns from file
RCLONE_EXCLUDE_IF_PRESENT=".rclone-ignore" # exclude directories if this filename is present
RCLONE_BWLIMIT="08:00,2M 00:00,off" # 2MB/s bandwidth limit between 8am and midnight
RCLONE_MIN_AGE=15m # skip sync for files created in the last 15 minutes
RCLONE_TRANSFERS=4 # number of file transfers to run in parallel
RCLONE_CHECKERS=8 # number of checkers to run in parallel
RCLONE_DELETE_EXCLUDED=true # delete files on dest excluded from sync
RCLONE_DRIVE_USE_TRASH=true # send all files to the trash when deleting files (Google Drive only)
RCLONE_IGNORE_CASE=true # ignore case when pattern matching 


### FUNCTIONS
#
# hash/obfuscate string
function hash() { md5sum < /dev/stdin | cut -f1 -d " "; }

# humanize seconds - https://unix.stackexchange.com/a/27014
function displaytime {
    local T=$1
    local D=$((T/60/60/24))
    local H=$((T/60/60%24))
    local M=$((T/60%60))
    local S=$((T%60))
    (( $D > 0 )) && printf '%dd' $D
    (( $H > 0 )) && printf '%dh' $H
    (( $M > 0 )) && printf '%dm' $M
    printf '%ds' $S
}

# takes unix epoch date as input and displays difference in seconds
function display_time_difference() {
    local seconds_diff="$(( $(date +%s) - $1 ))"
    echo "$(displaytime "$seconds_diff")"
}

# we'll use these to differentiate between executions and ensure only one sync happens at a time
src_dest_id="$(echo "${src}${dest}" | hash)"
execution_id="$(echo "$(date +%s)${src_dest_id}" | hash)"
lockfile="/tmp/rclone-${src_dest_id}.lock"

# let's get some help keeping our output formatted consistently
function format_output() {
    local timestamp="$(date -u +"%Y-%m-%dT%H:%M:%SZ")"
    local message="$(</dev/stdin)"
    if [ -n "$message" ]; then
        echo "${timestamp} | ${execution_id:0:7} | $message"
    fi
}


### VALIDATE
#
# check whether the path (local or remote) exists
function check_path() {
    rclone lsf "$1" >/dev/null 2>&1 || (echo "🚨 input path ($1) does not exist, script will exit" | format_output; exit 1)
}

check_path "$src"
check_path "$dest"


### LOGGING
#
# find somewhere local we can use for rclone's logging output
if [ -d "${src}" ]; then
    # src is local, dest is remote
    remote_name="$(echo "$dest" | cut -d ":" -f1)"
    default_log_path="${src}/.rclone/"
elif [ -d "${dest}" ]; then
    # src is remote, dest is local
    remote_name="$(echo "$src" | cut -d ":" -f1)"
    default_log_path="${dest}/.rclone/"
else
    remote_name="" # unknown remote
    default_log_path="/var/log/"
fi
log_file="${log_path:-$default_log_path}/rclone${remote_name:+-$remote_name}.log"
mkdir -p "$(dirname "$log_file")"


### RUN TIME
#
# function to run if there's already an active process running
exit_on_lock() {
    echo "🚨 another sync is already in progress, script will exit" | format_output
    exit 1
}

(
    # check if a lock file exists for this src/dest combo
    flock -n 9 || exit_on_lock

    # configure logging
    savelog -C -n -c 3 "$log_file" >/dev/null 2>&1

    # it's syncing time!
    echo "🏁 starting rclone sync ($src -> $dest)" | format_output
    start_time="$(date +%s)"
    rclone sync "$src" "$dest" -vv --log-file "$log_file"

    # finato
    duration="$(display_time_difference "$start_time")"
    echo "🎉 rclone sync complete! (took "$duration")" | format_output

) 9>"$lockfile"

Hey @ThomDietrich. I'm a fan of keeping things together where possible, but it's a pretty common convention to have config files located under ~/.config/<name>. It also means you're free to use rclone in other places (or run directly from the shell) without having to maintain or hunt down individual config files.

Your approach works wonderfully by exporting an environment variable though, no script modification necessary.

This looks good overall, but I have two related pieces of feedback. First, the example crontab is inconsistent with the built-in assignments of "src" and "dest":

#     */1 * * * * /home/johndoe/.config/rclone/rclone-cron.sh sync "/home/johndoe" "gdrive:" >> /var/log/rclone-cron.log 2>&1
...
src="${1}" # source
dest="${2}" # destination

It looks like the sync argument should just be removed, as it's not used and it offsets the other arguments and the script tries to upload a file called sync.

Second, the check_path function does not actually stop the script if it fails to find an input or output path. Perhaps this is specific to my bash (4.4.20, Ubuntu 18.04), but the exit code is 1 and yet the script continues. Here is a simple test routine to demonstrate, where /junk does not exist but /home does:

$ cat test.sh 
#!/bin/bash

# check whether the path (local or remote) exists
function check_path() {
   rclone lsf "$1" >/dev/null 2>&1 || (echo "path ($1) does not exist, exit"; exit 1)
}

echo "checking"

check_path "/junk"
echo $?
check_path "/home"
echo $?

echo "complete"
$ ./test.sh 
checking
path (/junk) does not exist, exit
1
0
complete

You can see that a bad path triggers a non-zero exit code from rclone, the error message, and a non-zero exit code. But the script still reaches "complete". This means that, if sync is provided as an argument the script will actually call rclone to sync the nonexistent folder. My solution for this is to eliminate the subshell from check_path, similar to how exit_on_lock is already done:

# check whether the path (local or remote) exists
function check_path() {
   if ! $(rclone lsf "$1" > /dev/null 2>&1); then
      echo "🚨 input path ($1) does not exist, script will exit" | format_output
      exit 1
   fi
}