zulip/scripts/purge-old-deployments

#!/usr/bin/env python3
import argparse
import os
import subprocess
import sys

ZULIP_PATH = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
sys.path.append(ZULIP_PATH)
from scripts.lib import clean_unused_caches
from scripts.lib.zulip_tools import (
    DEPLOYMENTS_DIR,
    get_recent_deployments,
    maybe_perform_purging,
    su_to_zulip,
)

LOCAL_GIT_CACHE_DIR = "/srv/zulip.git"


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description="This script can be used for cleaning old unused deployments.",
        epilog="Orphaned/unused caches older than threshold days will be automatically "
        "examined and removed.",
    )
    parser.add_argument(
        "--threshold",
        dest="threshold_days",
        type=int,
        default=14,
        metavar="<days>",
        help="Deployments older than threshold days will be deleted. (defaults to 14)",
    )
    parser.add_argument(
        "--dry-run",
        action="store_true",
        help="If specified then script will only print the deployments and "
        "caches that it will delete/keep back. It will not delete anything.",
    )
    parser.add_argument(
        "--verbose",
        action="store_true",
        help="If specified then script will print a detailed report of what is going on.",
    )
    parser.add_argument(
        "--no-print-headings",
        dest="no_headings",
        action="store_true",
        help="If specified then script will not print headings for "
        "what will be deleted/kept back.",
    )

    args = parser.parse_args()
    args.verbose |= args.dry_run  # Always print a detailed report in case of dry run.
    return args


def get_deployments_to_be_purged(recent_deployments: set[str]) -> set[str]:
    all_deployments = {
        os.path.join(DEPLOYMENTS_DIR, deployment) for deployment in os.listdir(DEPLOYMENTS_DIR)
    }
    deployments_to_purge = set()
    for deployment in all_deployments:
        # Deployments whose name is not in the format of a timestamp are
        # always included in the recent_deployments and are not deleted.
        if not os.path.isdir(deployment):
            # Skip things like uwsgi sockets.
            continue
        if not os.path.exists(os.path.join(deployment, "zerver")):
            # Skip things like "lock" that aren't actually a deployment directory
            continue
        if deployment not in recent_deployments:
            deployments_to_purge.add(deployment)
    return deployments_to_purge


def main() -> None:
    args = parse_args()
    deployments_to_keep = get_recent_deployments(args.threshold_days)
    deployments_to_purge = get_deployments_to_be_purged(deployments_to_keep)

    maybe_perform_purging(
        deployments_to_purge, deployments_to_keep, "deployment", args.dry_run, args.verbose, True
    )

    if not args.dry_run:
        if os.path.exists(LOCAL_GIT_CACHE_DIR):
            subprocess.check_call(
                ["git", "worktree", "prune"], cwd=LOCAL_GIT_CACHE_DIR, preexec_fn=su_to_zulip
            )

        if not args.no_headings:
            print("Deployments cleaned successfully...")
            print("Cleaning orphaned/unused caches...")

    # Call 'clean_unused_caches.py' script to clean any orphaned/unused caches.
    clean_unused_caches.main(args)
    print("Done!")


if __name__ == "__main__":
    main()