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