remove_hdp_snapshots.sh

#!/usr/bin/env bash

# Expecting 2 parameters.
# - Snapshoted Directory
# - Snapshots older then x days.


while [ $# -gt 0 ]; do
  case "$1" in
    --snapshot-dir)
      shift
      SNAPSHOT_DIR=$1
      shift
      ;;
    --older-than)
      shift
      OLDER_THAN=$1
      shift
      ;;
    *)
      break
      ;;
  esac
done

if [ "${SNAPSHOT_DIR}" == "" ]; then
  echo "Need to supply snapshot'ED' directory."
  echo "Usage: $0 --snapshot-dir <dir> --older-than <in-days>"
  exit -1
fi

if [ "${OLDER_THAN}" == "" ]; then
  echo "Need to supply older than days."
  echo "Usage: $0 --snapshot-dir <dir> --older-than <in-days>"
  exit -1
fi

declare -i NOW
NOW=`date +'%s'`*1000

declare -i OLDER_THAN_EPOCH
OLDER_THAN_EPOCH=${NOW}-${OLDER_THAN}*86400*1000

echo "Snapshot Directory: ${SNAPSHOT_DIR}"
echo "Older than (epoch in ms): ${OLDER_THAN_EPOCH}"

for i in `hdfs dfs -stat "%Y,%n" ${SNAPSHOT_DIR}/.snapshot/*`;do
    # Check if Snapshot is older than ...
    CUR_FILE_EPOCH=`echo "${i}" | cut -d "," -f 1`
    if [ ${CUR_FILE_EPOCH} \< ${OLDER_THAN_EPOCH} ]; then
        CUR_SNAPSHOT=`echo "${i}" | cut -d "," -f 2`
        echo "OLDER(epoch): ${CUR_FILE_EPOCH} vs ${OLDER_THAN_EPOCH} Snapshot:(dir) ${SNAPSHOT_DIR} - (snapshot)${CUR_SNAPSHOT}"
        echo "Removing snapshot: ${CUR_SNAPSHOT} from ${SNAPSHOT_DIR}"
        hdfs dfs -deleteSnapshot ${SNAPSHOT_DIR} ${CUR_SNAPSHOT}
    else
        echo "NEWER: ${CUR_FILE_EPOCH} - ${OLDER_THAN_EPOCH}"
    fi
done