Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 1 | #-*- coding: utf-8 -*- |
| 2 | |
| 3 | import datetime |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame] | 4 | import os |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 5 | |
| 6 | import utils |
| 7 | |
| 8 | from rsync_remote import RsyncRemote |
Max Rasskazov | 3eb1198 | 2015-07-16 19:45:17 +0300 | [diff] [blame^] | 9 | from utils import TimeStamp |
Max Rasskazov | d77a5e6 | 2015-06-19 17:47:30 +0300 | [diff] [blame] | 10 | |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 11 | |
Max Rasskazov | 3e83758 | 2015-06-17 18:44:15 +0300 | [diff] [blame] | 12 | class TRsync(RsyncRemote): |
Max Rasskazov | 6d7b5c8 | 2015-07-13 13:31:58 +0300 | [diff] [blame] | 13 | # TODO: possible check that rsync url is exists |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 14 | def __init__(self, |
| 15 | rsync_url, |
Max Rasskazov | 0127162 | 2015-06-17 02:35:09 +0300 | [diff] [blame] | 16 | snapshot_dir='snapshots', |
| 17 | latest_successful_postfix='latest', |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 18 | save_latest_days=14, |
| 19 | init_directory_structure=True, |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 20 | timestamp=None, |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 21 | ): |
Max Rasskazov | 3e83758 | 2015-06-17 18:44:15 +0300 | [diff] [blame] | 22 | super(TRsync, self).__init__(rsync_url) |
| 23 | self.logger = utils.logger.getChild('TRsync.' + rsync_url) |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 24 | self.timestamp = TimeStamp(timestamp) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 25 | self.logger.info('Using timestamp {}'.format(self.timestamp)) |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 26 | self.snapshot_dir = self.url.a_dir(snapshot_dir) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 27 | self.latest_successful_postfix = latest_successful_postfix |
| 28 | self.save_latest_days = save_latest_days |
| 29 | |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 30 | if init_directory_structure is True: |
| 31 | self.init_directory_structure() |
| 32 | |
| 33 | def init_directory_structure(self): |
Max Rasskazov | 0127162 | 2015-06-17 02:35:09 +0300 | [diff] [blame] | 34 | if self.url.url_type != 'path': |
| 35 | server_root = RsyncRemote(self.url.root) |
| 36 | return server_root.mkdir( |
| 37 | self.url.a_dir(self.url.path, self.snapshot_dir) |
| 38 | ) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 39 | |
Max Rasskazov | 501cea5 | 2015-06-29 18:11:25 +0300 | [diff] [blame] | 40 | def push(self, source, repo_name, symlinks=[], extra=None, save_diff=True): |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame] | 41 | repo_basename = os.path.split(repo_name)[-1] |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 42 | latest_path = self.url.a_file( |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 43 | self.snapshot_dir, |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame] | 44 | '{}-{}'.format(self.url.a_file(repo_basename), |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 45 | self.latest_successful_postfix) |
| 46 | ) |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 47 | |
| 48 | symlinks = list(symlinks) |
| 49 | symlinks.insert(0, latest_path) |
| 50 | |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 51 | snapshot_name = self.url.a_file( |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame] | 52 | '{}-{}'.format(self.url.a_file(repo_basename), self.timestamp) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 53 | ) |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 54 | repo_path = self.url.a_file(self.snapshot_dir, snapshot_name) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 55 | |
Max Rasskazov | 854399e | 2015-06-05 16:35:17 +0300 | [diff] [blame] | 56 | extra = '--link-dest={}'.format( |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 57 | self.url.a_file(self.url.path, latest_path) |
Max Rasskazov | 854399e | 2015-06-05 16:35:17 +0300 | [diff] [blame] | 58 | ) |
Max Rasskazov | b1b5083 | 2015-06-18 11:24:53 +0300 | [diff] [blame] | 59 | |
Max Rasskazov | b1b5083 | 2015-06-18 11:24:53 +0300 | [diff] [blame] | 60 | # TODO: split transaction run (push or pull), and |
| 61 | # commit/rollback functions. transaction must has possibility to |
| 62 | # rollback after commit for implementation of working with pool |
| 63 | # of servers. should be something like this: |
| 64 | # transactions = list() |
| 65 | # result = True |
| 66 | # for server in servers: |
| 67 | # transactions.append(server.push(source, repo_name)) |
| 68 | # result = result and transactions[-1].success |
| 69 | # if result is True: |
| 70 | # for transaction in transactions: |
| 71 | # transaction.commit() |
| 72 | # result = result and transactions[-1].success |
| 73 | # if result is False: |
| 74 | # for transaction in transactions: |
| 75 | # transaction.rollback() |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 76 | transaction = list() |
| 77 | try: |
| 78 | # start transaction |
Max Rasskazov | 3e83758 | 2015-06-17 18:44:15 +0300 | [diff] [blame] | 79 | result = super(TRsync, self).push(source, repo_path, extra) |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 80 | transaction.append(lambda p=repo_path: self.rmdir(p)) |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 81 | self.logger.info('{}'.format(result)) |
| 82 | |
Max Rasskazov | 501cea5 | 2015-06-29 18:11:25 +0300 | [diff] [blame] | 83 | if save_diff is True: |
| 84 | diff_file = self.tmp.get_file(content='{}'.format(result)) |
| 85 | diff_file_name = '{}.diff.txt'.format(repo_path) |
| 86 | super(TRsync, self).push(diff_file, diff_file_name, extra) |
| 87 | transaction.append(lambda f=diff_file_name: self.rmfile(f)) |
| 88 | self.logger.debug('Diff file {} created.' |
| 89 | ''.format(diff_file_name)) |
| 90 | |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 91 | for symlink in symlinks: |
| 92 | try: |
| 93 | tgt = [_[1] for _ in self.ls_symlinks(symlink)][0] |
| 94 | self.logger.info('Previous {} -> {}'.format(symlink, tgt)) |
| 95 | undo = lambda l=symlink, t=tgt: self.symlink(l, t) |
| 96 | except: |
| 97 | undo = lambda l=symlink: self.rmfile(l) |
| 98 | # TODO: implement detection of target relative symlink |
| 99 | if symlink.startswith(self.snapshot_dir): |
| 100 | self.symlink(symlink, snapshot_name) |
| 101 | else: |
| 102 | self.symlink(symlink, repo_path) |
| 103 | transaction.append(undo) |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 104 | |
Max Rasskazov | ad1518a | 2015-06-18 11:24:16 +0300 | [diff] [blame] | 105 | except RuntimeError: |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 106 | self.logger.error("Rollback transaction because some of sync" |
| 107 | "operation failed") |
| 108 | [_() for _ in reversed(transaction)] |
| 109 | raise |
| 110 | |
| 111 | try: |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 112 | # deleting of old snapshots ignored when assessing the transaction |
| 113 | # only warning |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 114 | self._remove_old_snapshots(repo_name) |
| 115 | except RuntimeError: |
| 116 | self.logger.warn("Old snapshots are not deleted. Ignore. " |
| 117 | "May be next time.") |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 118 | |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 119 | return result |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 120 | |
| 121 | def _remove_old_snapshots(self, repo_name, save_latest_days=None): |
| 122 | if save_latest_days is None: |
| 123 | save_latest_days = self.save_latest_days |
| 124 | if save_latest_days is None or save_latest_days is False: |
| 125 | # delete all snapshots |
| 126 | self.logger.info('Deletion all of the old snapshots ' |
| 127 | '(save_latest_days == {})' |
| 128 | ''.format(save_latest_days)) |
| 129 | save_latest_days = -1 |
| 130 | elif save_latest_days == 0: |
| 131 | # skipping deletion |
| 132 | self.logger.info('Skip deletion of old snapshots ' |
| 133 | '(save_latest_days == {})' |
| 134 | ''.format(save_latest_days)) |
| 135 | return |
| 136 | else: |
| 137 | # delete snapshots older than |
| 138 | self.logger.info('Deletion all of the unlinked snapshots older ' |
| 139 | 'than {0} days (save_latest_days == {0})' |
| 140 | ''.format(save_latest_days)) |
| 141 | warn_date = \ |
| 142 | self.timestamp.now - datetime.timedelta(days=save_latest_days) |
| 143 | warn_date = datetime.datetime.combine(warn_date, datetime.time(0)) |
| 144 | snapshots = self.ls_dirs( |
| 145 | self.url.a_dir(self.snapshot_dir), |
| 146 | pattern=r'^{}-{}$'.format( |
| 147 | repo_name, |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 148 | self.timestamp.snapshot_stamp_regexp |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 149 | ) |
| 150 | ) |
| 151 | links = self.ls_symlinks(self.url.a_dir()) |
| 152 | links += self.ls_symlinks(self.url.a_dir(self.snapshot_dir)) |
| 153 | for s in snapshots: |
| 154 | s_date = datetime.datetime.strptime( |
| 155 | s, |
| 156 | '{}-{}'.format(repo_name, |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 157 | self.timestamp.snapshot_stamp_format) |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 158 | ) |
| 159 | s_date = datetime.datetime.combine(s_date, datetime.time(0)) |
| 160 | s_path = self.url.a_dir(self.snapshot_dir, s) |
| 161 | if s_date < warn_date: |
| 162 | s_links = [_[0] for _ in links |
| 163 | if _[1] == s |
| 164 | or _[1].endswith('/{}'.format(s)) |
| 165 | ] |
| 166 | if not s_links: |
| 167 | self.rmdir(s_path) |
Max Rasskazov | 052e911 | 2015-06-29 18:43:15 +0300 | [diff] [blame] | 168 | self.rmfile(s_path + '.target.txt') |
| 169 | self.rmfile(s_path + '.diff.txt') |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 170 | else: |
| 171 | self.logger.info('Skip deletion of "{}" because there are ' |
| 172 | 'symlinks found: {}'.format(s, s_links)) |
| 173 | else: |
| 174 | self.logger.info('Skip deletion of "{}" because it newer than ' |
| 175 | '{} days'.format(s, save_latest_days)) |