zfs-snapshot-backup/copy_snapshots_to_remote.py

#!/usr/local/bin/python

# simple python module to copy incremental backups of zfs snapshots to a
# remote server

import pathlib
import subprocess


SSH_KEY_FILE = "/mnt/Datenspeicher/snap-backup-dataset/backup_key"
SSH_REMOTE = "zfs_snap_backup@etha.cpi.imtek.uni-freiburg.de"

REMOTE_PATH = "zfs-backups"
SCP_REMOTE_URL = f"{SSH_REMOTE}:~/{REMOTE_PATH}/"

ZFS_POOL = "Datenspeicher"
ZFS_ELAB_PREFIX = "elabfs-"

TMP_BACKUP_FOLDER = "/mnt/Datenspeicher/snap-backup-dataset/temporary-backups"


def call(arguments, as_text=False):
    """ run a command line argument

    simple wrapper around subprocess.run() with some sensible defaults

    :params arguments: list of command line arguments and parameters
    :params as_text: should the output treated as text
    :returns: bytesarray or string (if as_text is trueish)
    :raises subprocess.CalledProcessError: if command has not an exit value of 0
    """
    result = subprocess.run(
        arguments,
        check=True,
        stdout=subprocess.PIPE,
        universal_newlines=as_text,
    )
    return result.stdout

def get_member_name(filename):
    #elabfs-LukasMetzler@auto-20190809.0200-1w.gz
    if not filename.startswith(ZFS_ELAB_PREFIX):
        raise ValueError(f"not an elabfs snapshot: {filename}")
    prefix, rest = filename.split("@", 1)
    return prefix.replace(ZFS_ELAB_PREFIX, "")


def copy_snapshot(filepath):
    filename = filepath.name
    member = get_member_name(filename)
    print(f" - copying {filename}")
    remote_url = (
        f"{SSH_REMOTE}:~/{REMOTE_PATH}/{member}/{filename}"
    )
    copy_cmd = ["scp", "-i", SSH_KEY_FILE, str(filepath), remote_url]
    call(copy_cmd)


def batch_copy_snapshots():
    tmp_folder = pathlib.Path(TMP_BACKUP_FOLDER)
    gzip_items = (i for i in tmp_folder.iterdir() if i.suffix==".gz")
    for path in gzip_items:
        copy_snapshot(path)


if __name__ == "__main__":
    batch_copy_snapshots()