Delete oldest files in a folder until total used disk space is below a threshold

delete_oldest_until_threshold.sh

#!/usr/bin/env bash
#
# delete_oldest_until_threshold.sh
#
# Repeatedly deletes the oldest file in a folder until
# disk usage is below a specified threshold.
#
# Usage:
#   ./delete_oldest_until_threshold.sh -p /path/to/folder [-t threshold_in_percentage]
#
# Example:
#   ./delete_oldest_until_threshold.sh -p /var/log -t 85
#   ./delete_oldest_until_threshold.sh -p /mnt/data  # uses default threshold=80

######################################
# DEFAULTS
######################################
THRESHOLD=80

######################################
# PARSE ARGUMENTS
######################################
usage() {
  echo "Usage: $0 -p <path> [-t <threshold>]"
  echo "  -p PATH       : Path to the target directory (mandatory)."
  echo "  -t THRESHOLD  : Disk usage threshold (optional, default=$THRESHOLD)."
  exit 1
}

# Use getopts to handle -p and -t
while getopts ":p:t:" opt; do
  case $opt in
    p)
      TARGET_DIR="$OPTARG"
      ;;
    t)
      THRESHOLD="$OPTARG"
      ;;
    \?)
      echo "Invalid option: -$OPTARG" >&2
      usage
      ;;
    :)
      echo "Option -$OPTARG requires an argument." >&2
      usage
      ;;
  esac
done

# Ensure path is provided
if [ -z "$TARGET_DIR" ]; then
  echo "Error: Path (-p) is mandatory."
  usage
fi

######################################
# FUNCTIONS
######################################
get_usage() {
  # Return the usage of the partition containing $TARGET_DIR
  df --output=pcent "$TARGET_DIR" 2>/dev/null \
    | tail -n 1 \
    | tr -dc '0-9'
}

######################################
# MAIN SCRIPT
######################################

# 1) Get current disk usage
CURRENT_USAGE=$(get_usage)

if [ -z "$CURRENT_USAGE" ]; then
  echo "Error: Could not determine disk usage for '$TARGET_DIR'. Check that it exists."
  exit 1
fi

echo "Current usage: $CURRENT_USAGE%. Threshold: $THRESHOLD%."

# If we are already below threshold, no action needed
if [ "$CURRENT_USAGE" -lt "$THRESHOLD" ]; then
  echo "Disk usage is already below threshold. No action taken."
  exit 0
fi

# 2) Collect files in ascending order of modification time (oldest first)
#    -printf '%T@\t%p\n': prints epoch modification time + path
#    sort -k1,1n: sorts by numeric timestamp ascending
#    cut -f2-: removes the first column (timestamps), leaving just file paths
FILES=$(find "$TARGET_DIR" -type f -printf "%T@\t%p\n" 2>/dev/null \
        | sort -k1,1n \
        | cut -f2-)

# Convert the list of files into an array
IFS=$'\n' read -r -d '' -a FILE_ARRAY <<< "$(printf '%s\n' "$FILES")"

# 3) If no files found, exit
if [ "${#FILE_ARRAY[@]}" -eq 0 ]; then
  echo "No files found in '$TARGET_DIR' to delete."
  exit 0
fi

# 4) Delete files (oldest first) until usage is below threshold
for FILE in "${FILE_ARRAY[@]}"; do
  if [ -f "$FILE" ]; then
    echo "Deleting oldest file: $FILE"
    rm -f "$FILE"

    # Re-check usage after each deletion
    CURRENT_USAGE=$(get_usage)
    echo "Disk usage is now: $CURRENT_USAGE%."

    if [ "$CURRENT_USAGE" -lt "$THRESHOLD" ]; then
      echo "Disk usage is now below threshold ($THRESHOLD%). Done."
      exit 0
    fi
  fi
done

# 5) If we made it here, we removed all files in the target directory
#    but are still above threshold
echo "All files in '$TARGET_DIR' have been removed, yet usage is still $CURRENT_USAGE%."
echo "Further action might be necessary."
exit 0