summaryrefslogtreecommitdiff
path: root/modules/zfssnap/zfssnap.py
blob: d524f5dbb37418a2ca548a37de0a1044f756602b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
#!@python@/bin/python

import csv
import subprocess
import io
from distutils.util import strtobool 
from datetime import datetime, timezone, timedelta
from dateutil.tz import gettz, tzlocal
import pytimeparse
import argparse
import re

import sys

import logging

import shlex

from collections import defaultdict, OrderedDict, deque, namedtuple

import configparser
from xdg import BaseDirectory

from functools import cache

from math import floor

import asyncio


@cache
def _now():
    return datetime.now(timezone.utc)

def _snap_name(item, time=_now()):
    suffix = re.sub(r'\+00:00$', r'Z', time.isoformat(timespec='seconds'))
    return f'{item}@{suffix}'

def _log_cmd(*args):
    fmt_args = ' '.join(map(shlex.quote, args))
    logger.debug(f'Running command: {fmt_args}')

def _get_items():
    items = {}
    
    args = ['zfs', 'get', '-H', '-p', '-o', 'name,value', '-t', 'filesystem,volume', '-s', 'local,default,inherited,temporary,received', 'li.yggdrasil:auto-snapshot']
    _log_cmd(*args)
    with subprocess.Popen(args, stdout=subprocess.PIPE) as proc:
        text_stdout = io.TextIOWrapper(proc.stdout)
        reader = csv.reader(text_stdout, delimiter='\t', quoting=csv.QUOTE_NONE)
        Row = namedtuple('Row', ['name', 'setting'])
        for row in map(Row._make, reader):
            items[row.name] = bool(strtobool(row.setting))

    return items
        
def prune(config, dry_run, keep_newest):
    prunable_snapshots = set()
    args = ['zfs', 'get', '-H', '-p', '-o', 'name,value', '-t', 'snapshot', '-s', 'local', 'li.yggdrasil:is-auto-snapshot']
    _log_cmd(*args)
    with subprocess.Popen(args, stdout=subprocess.PIPE) as proc:
        text_stdout = io.TextIOWrapper(proc.stdout)
        reader = csv.reader(text_stdout, delimiter='\t', quoting=csv.QUOTE_NONE)
        Row = namedtuple('Row', ['name', 'is_auto_snapshot'])
        for row in map(Row._make, reader):
            if bool(strtobool(row.is_auto_snapshot)):
                prunable_snapshots.add(row.name)

    items = defaultdict(list)
    Snap = namedtuple('Snap', ['name', 'creation'])
    args = ['zfs', 'get', '-H', '-p', '-o', 'name,value', '-t', 'snapshot', 'creation']
    _log_cmd(*args)
    with subprocess.Popen(args, stdout=subprocess.PIPE) as proc:
        text_stdout = io.TextIOWrapper(proc.stdout)
        reader = csv.reader(text_stdout, delimiter='\t', quoting=csv.QUOTE_NONE)
        Row = namedtuple('Row', ['name', 'timestamp'])
        for row in map(Row._make, reader):
            if row.name not in prunable_snapshots:
                continue

            base_name, _, _ = row.name.rpartition('@')
            creation = datetime.fromtimestamp(int(row.timestamp), timezone.utc)
            items[base_name].append(Snap(name=row.name, creation=creation))

    kept_count = defaultdict(lambda: defaultdict(lambda: 0))
    KeptBecause = namedtuple('KeptBecause', ['rule', 'ix', 'base', 'period'])
    kept_because = OrderedDict()
    def keep_because(base, snap, rule, period=None):
        nonlocal KeptBecause, kept_count, kept_because
        kept_count[rule][base] += 1
        if snap not in kept_because:
            kept_because[snap] = deque()
        kept_because[snap].append(KeptBecause(rule=rule, ix=kept_count[rule][base], base=base, period=period))

    within = config.gettimedelta('KEEP', 'within')
    if within > timedelta(seconds=0):
        for base, snaps in items.items():
            time_ref = max(snaps, key=lambda snap: snap.creation, default=None)
            if not time_ref:
                logger.warn(f'Nothing to keep for ‘{base}’')
                continue

            logger.info(f'Using ‘{time_ref.name}’ as time reference for ‘{base}’')
            within_cutoff = time_ref.creation - within

            for snap in snaps:
                if snap.creation >= within_cutoff:
                    keep_because(base, snap.name, 'within')
    else:
        logger.warn('Skipping rule ‘within’ since retention period is zero')

    prune_timezone = config.gettimezone('KEEP', 'timezone', fallback=tzlocal)

    PRUNING_PATTERNS = OrderedDict([
        ("secondly", lambda t: t.strftime('%Y-%m-%d %H:%M:%S')),
        ("minutely", lambda t: t.strftime('%Y-%m-%d %H:%M')),
        ("5m", lambda t: (t.strftime('%Y-%m-%d %H'), floor(t.minute / 5) * 5)),
        ("hourly", lambda t: t.strftime('%Y-%m-%d %H')),
        ("daily", lambda t: t.strftime('%Y-%m-%d')),
        ("weekly", lambda t: t.strftime('%G-%V')),
        ("monthly", lambda t: t.strftime('%Y-%m')),
        ("yearly", lambda t: t.strftime('%Y')),
    ])

    for rule, pattern in PRUNING_PATTERNS.items():
        desired_count = config.getint('KEEP', rule, fallback=0)

        for base, snaps in items.items():
            periods = OrderedDict()
            
            for snap in sorted(snaps, key=lambda snap: snap.creation, reverse=keep_newest):
                period = pattern(snap.creation)
                if period not in periods:
                    periods[period] = deque()
                periods[period].append(snap)

            to_keep = desired_count
            ordered_periods = periods.items() if keep_newest else reversed(periods.items())
            for period, period_snaps in ordered_periods:
                if to_keep == 0:
                    break

                for snap in period_snaps:
                    keep_because(base, snap.name, rule, period=period)
                    to_keep -= 1
                    break

            if to_keep > 0:
                logger.debug(f'Missing {to_keep} to fulfill {rule}={desired_count} for ‘{base}’')

    for snap, reasons in kept_because.items():
        reasons_str = ', '.join(map(str, reasons))
        logger.info(f'Keeping ‘{snap}’ because: {reasons_str}')
    all_snaps = {snap.name for _, snaps in items.items() for snap in snaps}
    to_destroy = all_snaps - {*kept_because}
    if not to_destroy:
        logger.info('Nothing to prune')

    for snap in sorted(to_destroy):
        args = ['zfs', 'destroy']
        if dry_run:
            args += ['-n']
        args += [snap]
        _log_cmd(*args)
        subprocess.run(args, check=True)
        if dry_run:
            logger.info(f'Would have pruned ‘{snap}’')
        else:
            logger.info(f'Pruned ‘{snap}’')
    
def rename(snapshots, destroy=False, set_is_auto=False):
    args = ['zfs', 'get', '-H', '-p', '-o', 'name,value', 'creation', *snapshots]
    _log_cmd(*args)
    renamed_to = set()
    with subprocess.Popen(args, stdout=subprocess.PIPE) as proc:
        text_stdout = io.TextIOWrapper(proc.stdout)
        reader = csv.reader(text_stdout, delimiter='\t', quoting=csv.QUOTE_NONE)
        Row = namedtuple('Row', ['name', 'timestamp'])
        for row in map(Row._make, reader):
            creation = datetime.fromtimestamp(int(row.timestamp), timezone.utc)
            base_name, _, _ = row.name.rpartition('@')
            new_name = _snap_name(base_name, time=creation)
            if new_name == row.name:
                logger.debug(f'Not renaming ‘{row.name}’ since name is already correct')
                continue

            if new_name in renamed_to:
                if destroy:
                    logger.warning(f'Destroying ‘{row.name}’ since ‘{new_name}’ was already renamed to')
                    args = ['zfs', 'destroy', row.name]
                    _log_cmd(*args)
                    subprocess.run(args, check=True)
                else:
                    logger.info(f'Skipping ‘{row.name}’ since ‘{new_name}’ was already renamed to')

                continue

            logger.info(f'Renaming ‘{row.name}’ to ‘{new_name}’')
            args = ['zfs', 'rename', row.name, new_name]
            _log_cmd(*args)
            subprocess.run(args, check=True)
            renamed_to.add(new_name)

            if set_is_auto:
                logger.info(f'Setting is-auto-snapshot on ‘{new_name}’')
                args = ['zfs', 'set', 'li.yggdrasil:is-auto-snapshot=true', new_name]
                _log_cmd(*args)
                subprocess.run(args, check=True)

def autosnap():
    items = _get_items()

    recursive, single = set(), set()

    for item_name, is_included in items.items():
        if not is_included:
            continue

        children = {sub_name for sub_name in items if sub_name.startswith(f'{item_name}/')}
        is_recursive = all([items[sub_name] for sub_name in children])
        if is_recursive and children:
            recursive.add(item_name)
        else:
            single.add(item_name)

    for item_name in recursive | single:
        is_covered = any([item_name.startswith(f'{super_name}/') for super_name in recursive])
        if is_covered:
            try:
                recursive.remove(item_name)
            except KeyError:
                pass
            try:
                single.remove(item_name)
            except KeyError:
                pass

    all_snap_names = set()
    async def do_snapshot(*snap_items, recursive=False):
        nonlocal items, all_snap_names
        snap_names = {_snap_name(item) for item in snap_items}
        if recursive:
            for snap_item in snap_items:
                all_snap_names |= {_snap_name(item) for item in items if item.startswith(snap_item)}
        else:
            all_snap_names |= snap_names

        args = ['zfs', 'snapshot', '-o', 'li.yggdrasil:is-auto-snapshot=true']
        if recursive:
            args += ['-r']
        args += snap_names

        _log_cmd(*args)
        subprocess.run(args, check=True)

    tasks = []
    if single:
        tasks.append(do_snapshot(*single))
    if recursive:
        tasks.append(do_snapshot(*recursive, recursive=True))
    if not tasks:
        logger.warning('No snapshots to create')
    else:
        async def run_tasks():
            await asyncio.gather(*tasks)
        asyncio.run(run_tasks())
    for snap in all_snap_names:
        logger.info(f'Created ‘{snap}’')
    if all_snap_names:
        rename(snapshots=all_snap_names)

def main():
    global logger
    logger = logging.getLogger(__name__)
    console_handler = logging.StreamHandler()
    console_handler.setFormatter( logging.Formatter('[%(levelname)s](%(name)s): %(message)s') )
    if sys.stderr.isatty():
        console_handler.setFormatter( logging.Formatter('%(asctime)s [%(levelname)s](%(name)s): %(message)s') )
    logger.addHandler(console_handler)

    # log uncaught exceptions
    def log_exceptions(type, value, tb):
        global logger

        logger.error(value)
        sys.__excepthook__(type, value, tb) # calls default excepthook

    sys.excepthook = log_exceptions
    
    parser = argparse.ArgumentParser(prog='zfssnap')
    parser.add_argument('--verbose', '-v', action='count', default=0)
    subparsers = parser.add_subparsers()
    parser.set_defaults(cmd=autosnap)
    rename_parser = subparsers.add_parser('rename')
    rename_parser.add_argument('snapshots', nargs='+')
    rename_parser.add_argument('--destroy', action='store_true', default=False)
    rename_parser.add_argument('--set-is-auto', action='store_true', default=False)
    rename_parser.set_defaults(cmd=rename)
    prune_parser = subparsers.add_parser('prune')
    prune_parser.add_argument('--config', '-c', dest='config_files', nargs='*', default=list())
    prune_parser.add_argument('--dry-run', '-n', action='store_true', default=False)
    prune_parser.add_argument('--keep-newest', action='store_true', default=False)
    prune_parser.set_defaults(cmd=prune)
    args = parser.parse_args()

    if args.verbose <= 0:
        logger.setLevel(logging.WARNING)
    elif args.verbose <= 1:
        logger.setLevel(logging.INFO)
    else:
        logger.setLevel(logging.DEBUG)

    cmdArgs = {}
    for copy in {'snapshots', 'dry_run', 'destroy', 'keep_newest', 'set_is_auto'}:
        if copy in vars(args):
            cmdArgs[copy] = vars(args)[copy]
    if 'config_files' in vars(args):
        def convert_timedelta(secs_str):
            secs=pytimeparse.parse(secs_str)
            if secs is None:
                raise ValueError(f'Could not parse timedelta expression ‘{secs_str}’')
            return timedelta(seconds=secs)
        config = configparser.ConfigParser(converters={
            'timedelta': convert_timedelta,
            'timezone': gettz
        })
        search_files = args.config_files if args.config_files else [*BaseDirectory.load_config_paths('zfssnap.ini')]
        read_files = config.read(search_files)
        
        def format_config_files(files):
            if not files:
                return 'no files'
            return ', '.join(map(lambda file: f'‘{file}’', files))

        if not read_files:
            raise Exception(f'Found no config files. Tried: {format_config_files(search_files)}')

        logger.debug(f'Read following config files: {format_config_files(read_files)}')

        cmdArgs['config'] = config

    args.cmd(**cmdArgs)

sys.exit(main())