ArchiveInvalidator
5 years ago
ArchiveInvalidator.php
5 years ago
ArchivePurger.php
5 years ago
ArchiveQuery.php
5 years ago
ArchiveQueryFactory.php
5 years ago
Chunk.php
5 years ago
DataCollection.php
5 years ago
DataTableFactory.php
5 years ago
Parameters.php
5 years ago
ArchivePurger.php
331 lines
| 1 | <?php |
| 2 | /** |
| 3 | * Matomo - free/libre analytics platform |
| 4 | * |
| 5 | * @link https://matomo.org |
| 6 | * @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 7 | * |
| 8 | */ |
| 9 | namespace Piwik\Archive; |
| 10 | |
| 11 | use Piwik\ArchiveProcessor\Rules; |
| 12 | use Piwik\Config; |
| 13 | use Piwik\Container\StaticContainer; |
| 14 | use Piwik\DataAccess\ArchiveTableCreator; |
| 15 | use Piwik\DataAccess\Model; |
| 16 | use Piwik\Date; |
| 17 | use Piwik\Piwik; |
| 18 | use Psr\Log\LoggerInterface; |
| 19 | use Psr\Log\LogLevel; |
| 20 | |
| 21 | /** |
| 22 | * Service that purges temporary, error-ed, invalid and custom range archives from archive tables. |
| 23 | * |
| 24 | * Temporary archives are purged if they were archived before a specific time. The time is dependent |
| 25 | * on whether browser triggered archiving is enabled or not. |
| 26 | * |
| 27 | * Error-ed archives are purged w/o constraint. |
| 28 | * |
| 29 | * Invalid archives are purged if a new, valid, archive exists w/ the same site, date, period combination. |
| 30 | * Archives are marked as invalid via Piwik\Archive\ArchiveInvalidator. |
| 31 | */ |
| 32 | class ArchivePurger |
| 33 | { |
| 34 | /** |
| 35 | * @var Model |
| 36 | */ |
| 37 | private $model; |
| 38 | |
| 39 | /** |
| 40 | * Date threshold for purging custom range archives. Archives that are older than this date |
| 41 | * are purged unconditionally from the requested archive table. |
| 42 | * |
| 43 | * @var Date |
| 44 | */ |
| 45 | private $purgeCustomRangesOlderThan; |
| 46 | |
| 47 | /** |
| 48 | * Date to use for 'yesterday'. Exists so tests can override this value. |
| 49 | * |
| 50 | * @var Date |
| 51 | */ |
| 52 | private $yesterday; |
| 53 | |
| 54 | /** |
| 55 | * Date to use for 'today'. Exists so tests can override this value. |
| 56 | * |
| 57 | * @var $today |
| 58 | */ |
| 59 | private $today; |
| 60 | |
| 61 | /** |
| 62 | * Date to use for 'now'. Exists so tests can override this value. |
| 63 | * |
| 64 | * @var int |
| 65 | */ |
| 66 | private $now; |
| 67 | |
| 68 | /** |
| 69 | * @var LoggerInterface |
| 70 | */ |
| 71 | private $logger; |
| 72 | |
| 73 | public function __construct(Model $model = null, Date $purgeCustomRangesOlderThan = null, LoggerInterface $logger = null) |
| 74 | { |
| 75 | $this->model = $model ?: new Model(); |
| 76 | |
| 77 | $this->purgeCustomRangesOlderThan = $purgeCustomRangesOlderThan ?: self::getDefaultCustomRangeToPurgeAgeThreshold(); |
| 78 | |
| 79 | $this->yesterday = Date::factory('yesterday'); |
| 80 | $this->today = Date::factory('today'); |
| 81 | $this->now = time(); |
| 82 | $this->logger = $logger ?: StaticContainer::get('Psr\Log\LoggerInterface'); |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * Purge all invalidate archives for whom there are newer, valid archives from the archive |
| 87 | * table that stores data for `$date`. |
| 88 | * |
| 89 | * @param Date $date The date identifying the archive table. |
| 90 | * @return int The total number of archive rows deleted (from both the blog & numeric tables). |
| 91 | */ |
| 92 | public function purgeInvalidatedArchivesFrom(Date $date) |
| 93 | { |
| 94 | $numericTable = ArchiveTableCreator::getNumericTable($date); |
| 95 | |
| 96 | $archiveIds = $this->model->getInvalidatedArchiveIdsSafeToDelete($numericTable); |
| 97 | if (empty($archiveIds)) { |
| 98 | $this->logger->debug("No invalidated archives found in {table} with newer, valid archives.", array('table' => $numericTable)); |
| 99 | return 0; |
| 100 | } |
| 101 | |
| 102 | $emptyIdArchives = $this->model->getPlaceholderArchiveIds($numericTable); |
| 103 | $archiveIds = array_merge($archiveIds, $emptyIdArchives); |
| 104 | |
| 105 | $this->logger->info("Found {countArchiveIds} invalidated archives safe to delete in {table}.", array( |
| 106 | 'table' => $numericTable, 'countArchiveIds' => count($archiveIds) |
| 107 | )); |
| 108 | |
| 109 | $deletedRowCount = $this->deleteArchiveIds($date, $archiveIds); |
| 110 | |
| 111 | $this->logger->debug("Deleted {count} rows in {table} and its associated blob table.", array( |
| 112 | 'table' => $numericTable, 'count' => $deletedRowCount |
| 113 | )); |
| 114 | |
| 115 | return $deletedRowCount; |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Removes the outdated archives for the given month. |
| 120 | * (meaning they are marked with a done flag of ArchiveWriter::DONE_OK_TEMPORARY or ArchiveWriter::DONE_ERROR) |
| 121 | * |
| 122 | * @param Date $dateStart Only the month will be used |
| 123 | * @return int Returns the total number of rows deleted. |
| 124 | */ |
| 125 | public function purgeOutdatedArchives(Date $dateStart) |
| 126 | { |
| 127 | $purgeArchivesOlderThan = $this->getOldestTemporaryArchiveToKeepThreshold(); |
| 128 | $deletedRowCount = 0; |
| 129 | |
| 130 | $idArchivesToDelete = $this->getOutdatedArchiveIds($dateStart, $purgeArchivesOlderThan); |
| 131 | if (!empty($idArchivesToDelete)) { |
| 132 | $deletedRowCount = $this->deleteArchiveIds($dateStart, $idArchivesToDelete); |
| 133 | |
| 134 | $this->logger->info("Deleted {count} rows in archive tables (numeric + blob) for {date}.", array( |
| 135 | 'count' => $deletedRowCount, |
| 136 | 'date' => $dateStart |
| 137 | )); |
| 138 | } else { |
| 139 | $this->logger->debug("No outdated archives found in archive numeric table for {date}.", array('date' => $dateStart)); |
| 140 | } |
| 141 | |
| 142 | $this->logger->debug("Purging temporary archives: done [ purged archives older than {date} in {yearMonth} ] [Deleted IDs count: {deletedIds}]", array( |
| 143 | 'date' => $purgeArchivesOlderThan, |
| 144 | 'yearMonth' => $dateStart->toString('Y-m'), |
| 145 | 'deletedIds' => count($idArchivesToDelete), |
| 146 | )); |
| 147 | |
| 148 | return $deletedRowCount; |
| 149 | } |
| 150 | |
| 151 | public function purgeDeletedSiteArchives(Date $dateStart) |
| 152 | { |
| 153 | $archiveTable = ArchiveTableCreator::getNumericTable($dateStart); |
| 154 | $idArchivesToDelete = $this->model->getArchiveIdsForDeletedSites($archiveTable); |
| 155 | |
| 156 | return $this->purge($idArchivesToDelete, $dateStart, 'deleted sites'); |
| 157 | } |
| 158 | |
| 159 | /** |
| 160 | * @param Date $dateStart |
| 161 | * @param array $deletedSegments List of segments whose archives should be purged |
| 162 | * @return int |
| 163 | */ |
| 164 | public function purgeDeletedSegmentArchives(Date $dateStart, array $deletedSegments) |
| 165 | { |
| 166 | if (count($deletedSegments)) { |
| 167 | $idArchivesToDelete = $this->getDeletedSegmentArchiveIds($dateStart, $deletedSegments); |
| 168 | return $this->purge($idArchivesToDelete, $dateStart, 'deleted segments'); |
| 169 | } |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * Purge all numeric and blob archives with the given IDs from the database. |
| 174 | * @param array $idArchivesToDelete |
| 175 | * @param Date $dateStart |
| 176 | * @param string $reason |
| 177 | * @return int |
| 178 | */ |
| 179 | protected function purge(array $idArchivesToDelete, Date $dateStart, $reason) |
| 180 | { |
| 181 | $deletedRowCount = 0; |
| 182 | if (!empty($idArchivesToDelete)) { |
| 183 | $deletedRowCount = $this->deleteArchiveIds($dateStart, $idArchivesToDelete); |
| 184 | |
| 185 | $this->logger->info( |
| 186 | "Deleted {count} rows in archive tables (numeric + blob) for {reason} for {date}.", |
| 187 | array( |
| 188 | 'count' => $deletedRowCount, |
| 189 | 'date' => $dateStart, |
| 190 | 'reason' => $reason |
| 191 | ) |
| 192 | ); |
| 193 | |
| 194 | $this->logger->debug("[Deleted IDs count: {deletedIds}]", array( |
| 195 | 'deletedIds' => count($idArchivesToDelete), |
| 196 | )); |
| 197 | } else { |
| 198 | $this->logger->debug( |
| 199 | "No archives for {reason} found in archive numeric table for {date}.", |
| 200 | array('date' => $dateStart, 'reason' => $reason) |
| 201 | ); |
| 202 | } |
| 203 | |
| 204 | return $deletedRowCount; |
| 205 | } |
| 206 | |
| 207 | protected function getDeletedSegmentArchiveIds(Date $date, array $deletedSegments) |
| 208 | { |
| 209 | $archiveTable = ArchiveTableCreator::getNumericTable($date); |
| 210 | return $this->model->getArchiveIdsForSegments( |
| 211 | $archiveTable, $deletedSegments, $this->getOldestTemporaryArchiveToKeepThreshold() |
| 212 | ); |
| 213 | } |
| 214 | |
| 215 | protected function getOutdatedArchiveIds(Date $date, $purgeArchivesOlderThan) |
| 216 | { |
| 217 | $archiveTable = ArchiveTableCreator::getNumericTable($date); |
| 218 | |
| 219 | $result = $this->model->getTemporaryArchivesOlderThan($archiveTable, $purgeArchivesOlderThan); |
| 220 | |
| 221 | $idArchivesToDelete = array(); |
| 222 | if (!empty($result)) { |
| 223 | foreach ($result as $row) { |
| 224 | $idArchivesToDelete[] = $row['idarchive']; |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | return $idArchivesToDelete; |
| 229 | } |
| 230 | |
| 231 | /** |
| 232 | * Deleting "Custom Date Range" reports after 1 day, since they can be re-processed and would take up un-necessary space. |
| 233 | * |
| 234 | * @param $date Date |
| 235 | * @return int The total number of rows deleted from both the numeric & blob table. |
| 236 | */ |
| 237 | public function purgeArchivesWithPeriodRange(Date $date) |
| 238 | { |
| 239 | $numericTable = ArchiveTableCreator::getNumericTable($date); |
| 240 | $blobTable = ArchiveTableCreator::getBlobTable($date); |
| 241 | |
| 242 | $deletedCount = $this->model->deleteArchivesWithPeriod( |
| 243 | $numericTable, $blobTable, Piwik::$idPeriods['range'], $this->purgeCustomRangesOlderThan); |
| 244 | |
| 245 | $level = $deletedCount == 0 ? LogLevel::DEBUG : LogLevel::INFO; |
| 246 | $this->logger->log($level, "Purged {count} range archive rows from {numericTable} & {blobTable}.", array( |
| 247 | 'count' => $deletedCount, |
| 248 | 'numericTable' => $numericTable, |
| 249 | 'blobTable' => $blobTable |
| 250 | )); |
| 251 | |
| 252 | $this->logger->debug(" [ purged archives older than {threshold} ]", array('threshold' => $this->purgeCustomRangesOlderThan)); |
| 253 | |
| 254 | return $deletedCount; |
| 255 | } |
| 256 | |
| 257 | /** |
| 258 | * Deletes by batches Archive IDs in the specified month, |
| 259 | * |
| 260 | * @param Date $date |
| 261 | * @param $idArchivesToDelete |
| 262 | * @return int Number of rows deleted from both numeric + blob table. |
| 263 | */ |
| 264 | protected function deleteArchiveIds(Date $date, $idArchivesToDelete) |
| 265 | { |
| 266 | $batches = array_chunk($idArchivesToDelete, 1000); |
| 267 | $numericTable = ArchiveTableCreator::getNumericTable($date); |
| 268 | $blobTable = ArchiveTableCreator::getBlobTable($date); |
| 269 | |
| 270 | $deletedCount = 0; |
| 271 | foreach ($batches as $idsToDelete) { |
| 272 | $deletedCount += $this->model->deleteArchiveIds($numericTable, $blobTable, $idsToDelete); |
| 273 | } |
| 274 | return $deletedCount; |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * Returns a timestamp indicating outdated archives older than this timestamp (processed before) can be purged. |
| 279 | * |
| 280 | * @return int|bool Outdated archives older than this timestamp should be purged |
| 281 | */ |
| 282 | protected function getOldestTemporaryArchiveToKeepThreshold() |
| 283 | { |
| 284 | $temporaryArchivingTimeout = Rules::getTodayArchiveTimeToLive(); |
| 285 | if (Rules::isBrowserTriggerEnabled()) { |
| 286 | // If Browser Archiving is enabled, it is likely there are many more temporary archives |
| 287 | // We delete more often which is safe, since reports are re-processed on demand |
| 288 | return Date::factory($this->now - 2 * $temporaryArchivingTimeout)->getDateTime(); |
| 289 | } |
| 290 | |
| 291 | // If cron core:archive command is building the reports, we should keep all temporary reports from today |
| 292 | return $this->yesterday->getDateTime(); |
| 293 | } |
| 294 | |
| 295 | private static function getDefaultCustomRangeToPurgeAgeThreshold() |
| 296 | { |
| 297 | $daysRangesValid = Config::getInstance()->General['purge_date_range_archives_after_X_days']; |
| 298 | return Date::factory('today')->subDay($daysRangesValid)->getDateTime(); |
| 299 | } |
| 300 | |
| 301 | /** |
| 302 | * For tests. |
| 303 | * |
| 304 | * @param Date $yesterday |
| 305 | */ |
| 306 | public function setYesterdayDate(Date $yesterday) |
| 307 | { |
| 308 | $this->yesterday = $yesterday; |
| 309 | } |
| 310 | |
| 311 | /** |
| 312 | * For tests. |
| 313 | * |
| 314 | * @param Date $today |
| 315 | */ |
| 316 | public function setTodayDate(Date $today) |
| 317 | { |
| 318 | $this->today = $today; |
| 319 | } |
| 320 | |
| 321 | /** |
| 322 | * For tests. |
| 323 | * |
| 324 | * @param int $now |
| 325 | */ |
| 326 | public function setNow($now) |
| 327 | { |
| 328 | $this->now = $now; |
| 329 | } |
| 330 | } |
| 331 |