ansible-edda/playbooks/roles/backups/restic/files/restic-batch

#!/usr/bin/env python3

import argparse
import os
import subprocess
import yaml


def load_and_validate_config_dir(config_dir):
    if not os.path.isdir(config_dir):
        raise ValueError(f"{config_dir} is not a directory")

    return [
        load_and_validate_config_file(os.path.join(config_dir, file))
        for file in os.listdir(config_dir)
    ]


def load_and_validate_config_file(config_file_path):
    if not os.path.isfile(config_file_path):
        raise ValueError(f"{config_file_path} is not a file")

    with open(config_file_path, encoding="utf-8") as config_file:
        config = yaml.safe_load(config_file)

    for key in [
            "dataset",
            "aws_bucket_keys_file",
            "aws_bucket_endpoint",
            "aws_bucket_prefix",
            "restic_password_file",
            "restic_keep_daily",
            "restic_keep_monthly",
    ]:
        if key not in config:
            raise KeyError(f"{key} must be present in {config_file_path}")

    for file in [config["restic_password_file"], config["aws_bucket_keys_file"]]:
        if not os.path.isfile(file):
            raise ValueError(f"{file} is not a file")

    return config


def get_volume_datasets(root_dataset):
    zfs_list = subprocess.getoutput(
        f"zfs list -H -r {root_dataset} -o name,mountpoint"
    )
    zfs_list_lines = zfs_list.split('\n')
    zfs_list_lines_items = map(lambda l: l.split(), zfs_list_lines)
    return {
        os.path.relpath(dataset, root_dataset): {
            "dataset": dataset,
            "mountpoint": mountpoint,
        } for dataset, mountpoint in zfs_list_lines_items if os.path.ismount(mountpoint)
    }


def get_last_daily_snapshot(dataset):
    snapshots = subprocess.getoutput(
        f"zfs list -t snapshot -H -r {dataset} -o name -s creation"
    )
    daily_snapshots = filter(lambda s: s.endswith("_daily"), snapshots.split('\n'))
    last_daily_snapshot = list(daily_snapshots)[-1]
    assert '@' in last_daily_snapshot
    assert last_daily_snapshot.split('@')[0] == dataset
    return last_daily_snapshot


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Backup service data using restic")
    parser.add_argument("--config-dir", type=str, default="/etc/restic-batch.d",
                        help="Path to directory with YAML config files")
    args = parser.parse_args()

    config_list = load_and_validate_config_dir(args.config_dir)

    for config in config_list:
        volume_datasets = get_volume_datasets(config["dataset"])
        for volume, properties in volume_datasets.items():
            properties["snapshot"] = get_last_daily_snapshot(properties["dataset"])
        config["volume_datasets"] = volume_datasets

    for config in config_list:
        for volume, properties in config["volume_datasets"].items():
            snapshot = properties["snapshot"]

            mountpoint = properties["mountpoint"]
            backup_path = os.path.normpath(
                os.path.join("/", "mnt", os.path.relpath(mountpoint, "/"))
            )

            bucket_name = config["aws_bucket_prefix"]
            if volume != ".":
                bucket_name = f"{bucket_name}---{volume.replace('/', '---')}"
            bucket_repo = f"s3:https://{config['aws_bucket_endpoint']}/{bucket_name}"

            restic_cmd_base = [
                "/usr/local/bin/restic",
                "--password-file", config["restic_password_file"],
                "--repo", bucket_repo,
                "--option", "s3.storage-class=ONEZONE_IA",
            ]

            with open(config["aws_bucket_keys_file"], encoding="utf-8") as keys_file:
                aws_keys = yaml.safe_load(keys_file)
            environ = {**os.environ, **aws_keys}

            print(f"Backing up {bucket_name} : {snapshot}", flush=True)

            subprocess.run(["/usr/sbin/zfs", "clone",
                            "-o", f"mountpoint={backup_path}",
                            snapshot, "rpool/restic"],
                check=True,
            )
            try:
                try:
                    subprocess.run(restic_cmd_base + ["snapshots"], env=environ, check=True)
                except subprocess.CalledProcessError:
                    subprocess.run(restic_cmd_base + ["init"], env=environ, check=True)
                subprocess.run(restic_cmd_base + ["backup", "."],
                               cwd=backup_path, env=environ, check=True)
                subprocess.run(
                    restic_cmd_base + ["forget", "--prune",
                                       "--keep-daily", str(config["restic_keep_daily"]),
                                       "--keep-monthly", str(config["restic_keep_monthly"])],
                    env=environ, check=True,
                )
                subprocess.run(restic_cmd_base + ["check"], env=environ, check=True)
            finally:
                subprocess.run(["/usr/sbin/zfs", "destroy", "rpool/restic"], check=True)