src/Worker/DBClean.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Worker;
  23
  24 use Friendica\Core\Logger;
  25 use Friendica\Core\Worker;
  26 use Friendica\Database\DBA;
  27 use Friendica\DI;
  28
  29 /**
  30  * The script is called from time to time to clean the database entries and remove orphaned data.
  31  */
  32 class DBClean {
  33         public static function execute($stage = 0) {
  34
  35                 if (!DI::config()->get('system', 'dbclean', false)) {
  36                         return;
  37                 }
  38
  39                 if ($stage == 0) {
  40                         self::forkCleanProcess();
  41                 } else {
  42                         self::removeOrphans($stage);
  43                 }
  44         }
  45
  46         /**
  47          * Fork the different DBClean processes
  48          */
  49         private static function forkCleanProcess() {
  50                 // Get the expire days for step 8 and 9
  51                 $days = DI::config()->get('system', 'dbclean-expire-days', 0);
  52
  53                 for ($i = 1; $i <= 10; $i++) {
  54                         // Execute the background script for a step when it isn't finished.
  55                         // Execute step 8 and 9 only when $days is defined.
  56                         if (!DI::config()->get('system', 'finished-dbclean-'.$i, false) && (($i < 8) || ($i > 9) || ($days > 0))) {
  57                                 Worker::add(PRIORITY_LOW, 'DBClean', $i);
  58                         }
  59                 }
  60         }
  61
  62         /**
  63          * Remove orphaned database entries
  64          *
  65          * @param integer $stage What should be deleted?
  66          *
  67          * Values for $stage:
  68          * ------------------
  69          *  1:    Old global item entries from item table without user copy.
  70          *  2:    Items without parents.
  71          *  3:    Orphaned data from thread table.
  72          *  4:    Orphaned data from notify table.
  73          *  5:    Orphaned data from notify-threads table.
  74          *  6:    Legacy functionality (removed)
  75          *  7:    Orphaned data from term table.
  76          *  8:    Expired threads.
  77          *  9:    Old global item entries from expired threads.
  78          * 10:    Old conversations.
  79          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  80          */
  81         private static function removeOrphans($stage) {
  82                 // We split the deletion in many small tasks
  83                 $limit = DI::config()->get('system', 'dbclean-expire-limit', 1000);
  84
  85                 // Get the expire days for step 8 and 9
  86                 $days = DI::config()->get('system', 'dbclean-expire-days', 0);
  87                 $days_unclaimed = DI::config()->get('system', 'dbclean-expire-unclaimed', 90);
  88
  89                 if ($days_unclaimed == 0) {
  90                         $days_unclaimed = $days;
  91                 }
  92
  93                 if ($stage == 1) {
  94                         if ($days_unclaimed <= 0) {
  95                                 return;
  96                         }
  97
  98                         $last_id = DI::config()->get('system', 'dbclean-last-id-1', 0);
  99
 100                         Logger::log("Deleting old global item entries from item table without user copy. Last ID: ".$last_id);
 101                         $r = DBA::p("SELECT `id`, `guid` FROM `item` WHERE `uid` = 0 AND
 102                                                 NOT EXISTS (SELECT `guid` FROM `item` AS `i` WHERE `item`.`guid` = `i`.`guid` AND `i`.`uid` != 0) AND
 103                                                 `received` < UTC_TIMESTAMP() - INTERVAL ? DAY AND `id` >= ?
 104                                         ORDER BY `id` LIMIT ?", $days_unclaimed, $last_id, $limit);
 105                         $count = DBA::numRows($r);
 106                         if ($count > 0) {
 107                                 Logger::log("found global item orphans: ".$count);
 108                                 while ($orphan = DBA::fetch($r)) {
 109                                         $last_id = $orphan["id"];
 110                                         Logger::info('Delete global orphan item', ['id' => $orphan['id'], 'guid' => $orphan['guid']]);
 111                                         DBA::delete('item', ['id' => $orphan["id"]]);
 112                                 }
 113                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 1, $last_id);
 114                         } else {
 115                                 Logger::log("No global item orphans found");
 116                         }
 117                         DBA::close($r);
 118                         Logger::log("Done deleting ".$count." old global item entries from item table without user copy. Last ID: ".$last_id);
 119
 120                         DI::config()->set('system', 'dbclean-last-id-1', $last_id);
 121                 } elseif ($stage == 2) {
 122                         $last_id = DI::config()->get('system', 'dbclean-last-id-2', 0);
 123
 124                         Logger::log("Deleting items without parents. Last ID: ".$last_id);
 125                         $r = DBA::p("SELECT `id`, `guid` FROM `item`
 126                                         WHERE NOT EXISTS (SELECT `id` FROM `item` AS `i` WHERE `item`.`parent` = `i`.`id`)
 127                                         AND `id` >= ? ORDER BY `id` LIMIT ?", $last_id, $limit);
 128                         $count = DBA::numRows($r);
 129                         if ($count > 0) {
 130                                 Logger::log("found item orphans without parents: ".$count);
 131                                 while ($orphan = DBA::fetch($r)) {
 132                                         $last_id = $orphan["id"];
 133                                         Logger::info('Delete orphan item', ['id' => $orphan['id'], 'guid' => $orphan['guid']]);
 134                                         DBA::delete('item', ['id' => $orphan["id"]]);
 135                                 }
 136                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 2, $last_id);
 137                         } else {
 138                                 Logger::log("No item orphans without parents found");
 139                         }
 140                         DBA::close($r);
 141                         Logger::log("Done deleting ".$count." items without parents. Last ID: ".$last_id);
 142
 143                         DI::config()->set('system', 'dbclean-last-id-2', $last_id);
 144
 145                         if ($count < $limit) {
 146                                 DI::config()->set('system', 'finished-dbclean-2', true);
 147                         }
 148                 } elseif ($stage == 3) {
 149                         $last_id = DI::config()->get('system', 'dbclean-last-id-3', 0);
 150
 151                         Logger::log("Deleting orphaned data from thread table. Last ID: ".$last_id);
 152                         $r = DBA::p("SELECT `iid` FROM `thread`
 153                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`parent` = `thread`.`iid`) AND `iid` >= ?
 154                                         ORDER BY `iid` LIMIT ?", $last_id, $limit);
 155                         $count = DBA::numRows($r);
 156                         if ($count > 0) {
 157                                 Logger::log("found thread orphans: ".$count);
 158                                 while ($orphan = DBA::fetch($r)) {
 159                                         $last_id = $orphan["iid"];
 160                                         DBA::delete('thread', ['iid' => $orphan["iid"]]);
 161                                 }
 162                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 3, $last_id);
 163                         } else {
 164                                 Logger::log("No thread orphans found");
 165                         }
 166                         DBA::close($r);
 167                         Logger::log("Done deleting ".$count." orphaned data from thread table. Last ID: ".$last_id);
 168
 169                         DI::config()->set('system', 'dbclean-last-id-3', $last_id);
 170
 171                         if ($count < $limit) {
 172                                 DI::config()->set('system', 'finished-dbclean-3', true);
 173                         }
 174                 } elseif ($stage == 4) {
 175                         $last_id = DI::config()->get('system', 'dbclean-last-id-4', 0);
 176
 177                         Logger::log("Deleting orphaned data from notify table. Last ID: ".$last_id);
 178                         $r = DBA::p("SELECT `iid`, `id` FROM `notify`
 179                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`id` = `notify`.`iid`) AND `id` >= ?
 180                                         ORDER BY `id` LIMIT ?", $last_id, $limit);
 181                         $count = DBA::numRows($r);
 182                         if ($count > 0) {
 183                                 Logger::log("found notify orphans: ".$count);
 184                                 while ($orphan = DBA::fetch($r)) {
 185                                         $last_id = $orphan["id"];
 186                                         DBA::delete('notify', ['iid' => $orphan["iid"]]);
 187                                 }
 188                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 4, $last_id);
 189                         } else {
 190                                 Logger::log("No notify orphans found");
 191                         }
 192                         DBA::close($r);
 193                         Logger::log("Done deleting ".$count." orphaned data from notify table. Last ID: ".$last_id);
 194
 195                         DI::config()->set('system', 'dbclean-last-id-4', $last_id);
 196
 197                         if ($count < $limit) {
 198                                 DI::config()->set('system', 'finished-dbclean-4', true);
 199                         }
 200                 } elseif ($stage == 5) {
 201                         $last_id = DI::config()->get('system', 'dbclean-last-id-5', 0);
 202
 203                         Logger::log("Deleting orphaned data from notify-threads table. Last ID: ".$last_id);
 204                         $r = DBA::p("SELECT `id` FROM `notify-threads`
 205                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`parent` = `notify-threads`.`master-parent-item`) AND `id` >= ?
 206                                         ORDER BY `id` LIMIT ?", $last_id, $limit);
 207                         $count = DBA::numRows($r);
 208                         if ($count > 0) {
 209                                 Logger::log("found notify-threads orphans: ".$count);
 210                                 while ($orphan = DBA::fetch($r)) {
 211                                         $last_id = $orphan["id"];
 212                                         DBA::delete('notify-threads', ['id' => $orphan["id"]]);
 213                                 }
 214                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 5, $last_id);
 215                         } else {
 216                                 Logger::log("No notify-threads orphans found");
 217                         }
 218                         DBA::close($r);
 219                         Logger::log("Done deleting ".$count." orphaned data from notify-threads table. Last ID: ".$last_id);
 220
 221                         DI::config()->set('system', 'dbclean-last-id-5', $last_id);
 222
 223                         if ($count < $limit) {
 224                                 DI::config()->set('system', 'finished-dbclean-5', true);
 225                         }
 226                 } elseif ($stage == 6) {
 227                         // The legacy functionality had been removed
 228                         DI::config()->set('system', 'finished-dbclean-6', true);
 229                 } elseif ($stage == 7) {
 230                         $last_id = DI::config()->get('system', 'dbclean-last-id-7', 0);
 231
 232                         Logger::log("Deleting orphaned data from term table. Last ID: ".$last_id);
 233                         $r = DBA::p("SELECT `oid`, `tid` FROM `term`
 234                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`id` = `term`.`oid`) AND `tid` >= ?
 235                                         ORDER BY `tid` LIMIT ?", $last_id, $limit);
 236                         $count = DBA::numRows($r);
 237                         if ($count > 0) {
 238                                 Logger::log("found term orphans: ".$count);
 239                                 while ($orphan = DBA::fetch($r)) {
 240                                         $last_id = $orphan["tid"];
 241                                         DBA::delete('term', ['oid' => $orphan["oid"]]);
 242                                 }
 243                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 7, $last_id);
 244                         } else {
 245                                 Logger::log("No term orphans found");
 246                         }
 247                         DBA::close($r);
 248                         Logger::log("Done deleting ".$count." orphaned data from term table. Last ID: ".$last_id);
 249
 250                         DI::config()->set('system', 'dbclean-last-id-7', $last_id);
 251
 252                         if ($count < $limit) {
 253                                 DI::config()->set('system', 'finished-dbclean-7', true);
 254                         }
 255                 } elseif ($stage == 8) {
 256                         if ($days <= 0) {
 257                                 return;
 258                         }
 259
 260                         $last_id = DI::config()->get('system', 'dbclean-last-id-8', 0);
 261
 262                         Logger::log("Deleting expired threads. Last ID: ".$last_id);
 263                         $r = DBA::p("SELECT `thread`.`iid` FROM `thread`
 264                                         INNER JOIN `contact` ON `thread`.`contact-id` = `contact`.`id` AND NOT `notify_new_posts`
 265                                         WHERE `thread`.`received` < UTC_TIMESTAMP() - INTERVAL ? DAY
 266                                                 AND NOT `thread`.`mention` AND NOT `thread`.`starred`
 267                                                 AND NOT `thread`.`wall` AND NOT `thread`.`origin`
 268                                                 AND `thread`.`uid` != 0 AND `thread`.`iid` >= ?
 269                                                 AND NOT `thread`.`iid` IN (SELECT `parent` FROM `item`
 270                                                                 WHERE (`item`.`starred` OR (`item`.`resource-id` != '')
 271                                                                         OR (`item`.`file` != '') OR (`item`.`event-id` != '')
 272                                                                         OR (`item`.`attach` != '') OR `item`.`wall` OR `item`.`origin`)
 273                                                                         AND `item`.`parent` = `thread`.`iid`)
 274                                         ORDER BY `thread`.`iid` LIMIT ?", $days, $last_id, $limit);
 275                         $count = DBA::numRows($r);
 276                         if ($count > 0) {
 277                                 Logger::log("found expired threads: ".$count);
 278                                 while ($thread = DBA::fetch($r)) {
 279                                         $last_id = $thread["iid"];
 280                                         DBA::delete('thread', ['iid' => $thread["iid"]]);
 281                                 }
 282                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 8, $last_id);
 283                         } else {
 284                                 Logger::log("No expired threads found");
 285                         }
 286                         DBA::close($r);
 287                         Logger::log("Done deleting ".$count." expired threads. Last ID: ".$last_id);
 288
 289                         DI::config()->set('system', 'dbclean-last-id-8', $last_id);
 290                 } elseif ($stage == 9) {
 291                         if ($days <= 0) {
 292                                 return;
 293                         }
 294
 295                         $last_id = DI::config()->get('system', 'dbclean-last-id-9', 0);
 296                         $till_id = DI::config()->get('system', 'dbclean-last-id-8', 0);
 297
 298                         Logger::log("Deleting old global item entries from expired threads from ID ".$last_id." to ID ".$till_id);
 299                         $r = DBA::p("SELECT `id`, `guid` FROM `item` WHERE `uid` = 0 AND
 300                                                 NOT EXISTS (SELECT `guid` FROM `item` AS `i` WHERE `item`.`guid` = `i`.`guid` AND `i`.`uid` != 0) AND
 301                                                 `received` < UTC_TIMESTAMP() - INTERVAL 90 DAY AND `id` >= ? AND `id` <= ?
 302                                         ORDER BY `id` LIMIT ?", $last_id, $till_id, $limit);
 303                         $count = DBA::numRows($r);
 304                         if ($count > 0) {
 305                                 Logger::log("found global item entries from expired threads: ".$count);
 306                                 while ($orphan = DBA::fetch($r)) {
 307                                         $last_id = $orphan["id"];
 308                                         Logger::info('Delete expired thread item', ['id' => $orphan['id'], 'guid' => $orphan['guid']]);
 309                                         DBA::delete('item', ['id' => $orphan["id"]]);
 310                                 }
 311                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 9, $last_id);
 312                         } else {
 313                                 Logger::log("No global item entries from expired threads");
 314                         }
 315                         DBA::close($r);
 316                         Logger::log("Done deleting ".$count." old global item entries from expired threads. Last ID: ".$last_id);
 317
 318                         DI::config()->set('system', 'dbclean-last-id-9', $last_id);
 319                 } elseif ($stage == 10) {
 320                         $last_id = DI::config()->get('system', 'dbclean-last-id-10', 0);
 321                         $days = intval(DI::config()->get('system', 'dbclean_expire_conversation', 90));
 322
 323                         Logger::log("Deleting old conversations. Last created: ".$last_id);
 324                         $r = DBA::p("SELECT `received`, `item-uri` FROM `conversation`
 325                                         WHERE `received` < UTC_TIMESTAMP() - INTERVAL ? DAY
 326                                         ORDER BY `received` LIMIT ?", $days, $limit);
 327                         $count = DBA::numRows($r);
 328                         if ($count > 0) {
 329                                 Logger::log("found old conversations: ".$count);
 330                                 while ($orphan = DBA::fetch($r)) {
 331                                         $last_id = $orphan["received"];
 332                                         DBA::delete('conversation', ['item-uri' => $orphan["item-uri"]]);
 333                                 }
 334                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 10, $last_id);
 335                         } else {
 336                                 Logger::log("No old conversations found");
 337                         }
 338                         DBA::close($r);
 339                         Logger::log("Done deleting ".$count." conversations. Last created: ".$last_id);
 340
 341                         DI::config()->set('system', 'dbclean-last-id-10', $last_id);
 342                 }
 343         }
 344 }