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 $stage:
  68          * ------------------
  69          *  1:    Old global item entries from item table without user copy.
  70          *  2:    Items without parents.
  71          *  3:    Orphaned data from thread table.
  72          *  4:    Orphaned data from notify table.
  73          *  5:    Orphaned data from notify-threads table.
  74          *  6:    Orphaned data from sign table.
  75          *  7:    Orphaned data from term table.
  76          *  8:    Expired threads.
  77          *  9:    Old global item entries from expired threads.
  78          * 10:    Old conversations.
  79          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  80          */
  81         private static function removeOrphans($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` 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                                         DBA::delete('item', ['id' => $orphan["id"]]);
 111                                 }
 112                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 1, $last_id);
 113                         } else {
 114                                 Logger::log("No global item orphans found");
 115                         }
 116                         DBA::close($r);
 117                         Logger::log("Done deleting ".$count." old global item entries from item table without user copy. Last ID: ".$last_id);
 118
 119                         DI::config()->set('system', 'dbclean-last-id-1', $last_id);
 120                 } elseif ($stage == 2) {
 121                         $last_id = DI::config()->get('system', 'dbclean-last-id-2', 0);
 122
 123                         Logger::log("Deleting items without parents. Last ID: ".$last_id);
 124                         $r = DBA::p("SELECT `id` FROM `item`
 125                                         WHERE NOT EXISTS (SELECT `id` FROM `item` AS `i` WHERE `item`.`parent` = `i`.`id`)
 126                                         AND `id` >= ? ORDER BY `id` LIMIT ?", $last_id, $limit);
 127                         $count = DBA::numRows($r);
 128                         if ($count > 0) {
 129                                 Logger::log("found item orphans without parents: ".$count);
 130                                 while ($orphan = DBA::fetch($r)) {
 131                                         $last_id = $orphan["id"];
 132                                         DBA::delete('item', ['id' => $orphan["id"]]);
 133                                 }
 134                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 2, $last_id);
 135                         } else {
 136                                 Logger::log("No item orphans without parents found");
 137                         }
 138                         DBA::close($r);
 139                         Logger::log("Done deleting ".$count." items without parents. Last ID: ".$last_id);
 140
 141                         DI::config()->set('system', 'dbclean-last-id-2', $last_id);
 142
 143                         if ($count < $limit) {
 144                                 DI::config()->set('system', 'finished-dbclean-2', true);
 145                         }
 146                 } elseif ($stage == 3) {
 147                         $last_id = DI::config()->get('system', 'dbclean-last-id-3', 0);
 148
 149                         Logger::log("Deleting orphaned data from thread table. Last ID: ".$last_id);
 150                         $r = DBA::p("SELECT `iid` FROM `thread`
 151                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`parent` = `thread`.`iid`) AND `iid` >= ?
 152                                         ORDER BY `iid` LIMIT ?", $last_id, $limit);
 153                         $count = DBA::numRows($r);
 154                         if ($count > 0) {
 155                                 Logger::log("found thread orphans: ".$count);
 156                                 while ($orphan = DBA::fetch($r)) {
 157                                         $last_id = $orphan["iid"];
 158                                         DBA::delete('thread', ['iid' => $orphan["iid"]]);
 159                                 }
 160                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 3, $last_id);
 161                         } else {
 162                                 Logger::log("No thread orphans found");
 163                         }
 164                         DBA::close($r);
 165                         Logger::log("Done deleting ".$count." orphaned data from thread table. Last ID: ".$last_id);
 166
 167                         DI::config()->set('system', 'dbclean-last-id-3', $last_id);
 168
 169                         if ($count < $limit) {
 170                                 DI::config()->set('system', 'finished-dbclean-3', true);
 171                         }
 172                 } elseif ($stage == 4) {
 173                         $last_id = DI::config()->get('system', 'dbclean-last-id-4', 0);
 174
 175                         Logger::log("Deleting orphaned data from notify table. Last ID: ".$last_id);
 176                         $r = DBA::p("SELECT `iid`, `id` FROM `notify`
 177                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`id` = `notify`.`iid`) AND `id` >= ?
 178                                         ORDER BY `id` LIMIT ?", $last_id, $limit);
 179                         $count = DBA::numRows($r);
 180                         if ($count > 0) {
 181                                 Logger::log("found notify orphans: ".$count);
 182                                 while ($orphan = DBA::fetch($r)) {
 183                                         $last_id = $orphan["id"];
 184                                         DBA::delete('notify', ['iid' => $orphan["iid"]]);
 185                                 }
 186                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 4, $last_id);
 187                         } else {
 188                                 Logger::log("No notify orphans found");
 189                         }
 190                         DBA::close($r);
 191                         Logger::log("Done deleting ".$count." orphaned data from notify table. Last ID: ".$last_id);
 192
 193                         DI::config()->set('system', 'dbclean-last-id-4', $last_id);
 194
 195                         if ($count < $limit) {
 196                                 DI::config()->set('system', 'finished-dbclean-4', true);
 197                         }
 198                 } elseif ($stage == 5) {
 199                         $last_id = DI::config()->get('system', 'dbclean-last-id-5', 0);
 200
 201                         Logger::log("Deleting orphaned data from notify-threads table. Last ID: ".$last_id);
 202                         $r = DBA::p("SELECT `id` FROM `notify-threads`
 203                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`parent` = `notify-threads`.`master-parent-item`) AND `id` >= ?
 204                                         ORDER BY `id` LIMIT ?", $last_id, $limit);
 205                         $count = DBA::numRows($r);
 206                         if ($count > 0) {
 207                                 Logger::log("found notify-threads orphans: ".$count);
 208                                 while ($orphan = DBA::fetch($r)) {
 209                                         $last_id = $orphan["id"];
 210                                         DBA::delete('notify-threads', ['id' => $orphan["id"]]);
 211                                 }
 212                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 5, $last_id);
 213                         } else {
 214                                 Logger::log("No notify-threads orphans found");
 215                         }
 216                         DBA::close($r);
 217                         Logger::log("Done deleting ".$count." orphaned data from notify-threads table. Last ID: ".$last_id);
 218
 219                         DI::config()->set('system', 'dbclean-last-id-5', $last_id);
 220
 221                         if ($count < $limit) {
 222                                 DI::config()->set('system', 'finished-dbclean-5', true);
 223                         }
 224                 } elseif ($stage == 6) {
 225                         $last_id = DI::config()->get('system', 'dbclean-last-id-6', 0);
 226
 227                         Logger::log("Deleting orphaned data from sign table. Last ID: ".$last_id);
 228                         $r = DBA::p("SELECT `iid`, `id` FROM `sign`
 229                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`id` = `sign`.`iid`) AND `id` >= ?
 230                                         ORDER BY `id` LIMIT ?", $last_id, $limit);
 231                         $count = DBA::numRows($r);
 232                         if ($count > 0) {
 233                                 Logger::log("found sign orphans: ".$count);
 234                                 while ($orphan = DBA::fetch($r)) {
 235                                         $last_id = $orphan["id"];
 236                                         DBA::delete('sign', ['iid' => $orphan["iid"]]);
 237                                 }
 238                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 6, $last_id);
 239                         } else {
 240                                 Logger::log("No sign orphans found");
 241                         }
 242                         DBA::close($r);
 243                         Logger::log("Done deleting ".$count." orphaned data from sign table. Last ID: ".$last_id);
 244
 245                         DI::config()->set('system', 'dbclean-last-id-6', $last_id);
 246
 247                         if ($count < $limit) {
 248                                 DI::config()->set('system', 'finished-dbclean-6', true);
 249                         }
 250                 } elseif ($stage == 7) {
 251                         $last_id = DI::config()->get('system', 'dbclean-last-id-7', 0);
 252
 253                         Logger::log("Deleting orphaned data from term table. Last ID: ".$last_id);
 254                         $r = DBA::p("SELECT `oid`, `tid` FROM `term`
 255                                         WHERE NOT EXISTS (SELECT `id` FROM `item` WHERE `item`.`id` = `term`.`oid`) AND `tid` >= ?
 256                                         ORDER BY `tid` LIMIT ?", $last_id, $limit);
 257                         $count = DBA::numRows($r);
 258                         if ($count > 0) {
 259                                 Logger::log("found term orphans: ".$count);
 260                                 while ($orphan = DBA::fetch($r)) {
 261                                         $last_id = $orphan["tid"];
 262                                         DBA::delete('term', ['oid' => $orphan["oid"]]);
 263                                 }
 264                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 7, $last_id);
 265                         } else {
 266                                 Logger::log("No term orphans found");
 267                         }
 268                         DBA::close($r);
 269                         Logger::log("Done deleting ".$count." orphaned data from term table. Last ID: ".$last_id);
 270
 271                         DI::config()->set('system', 'dbclean-last-id-7', $last_id);
 272
 273                         if ($count < $limit) {
 274                                 DI::config()->set('system', 'finished-dbclean-7', true);
 275                         }
 276                 } elseif ($stage == 8) {
 277                         if ($days <= 0) {
 278                                 return;
 279                         }
 280
 281                         $last_id = DI::config()->get('system', 'dbclean-last-id-8', 0);
 282
 283                         Logger::log("Deleting expired threads. Last ID: ".$last_id);
 284                         $r = DBA::p("SELECT `thread`.`iid` FROM `thread`
 285                                         INNER JOIN `contact` ON `thread`.`contact-id` = `contact`.`id` AND NOT `notify_new_posts`
 286                                         WHERE `thread`.`received` < UTC_TIMESTAMP() - INTERVAL ? DAY
 287                                                 AND NOT `thread`.`mention` AND NOT `thread`.`starred`
 288                                                 AND NOT `thread`.`wall` AND NOT `thread`.`origin`
 289                                                 AND `thread`.`uid` != 0 AND `thread`.`iid` >= ?
 290                                                 AND NOT `thread`.`iid` IN (SELECT `parent` FROM `item`
 291                                                                 WHERE (`item`.`starred` OR (`item`.`resource-id` != '')
 292                                                                         OR (`item`.`file` != '') OR (`item`.`event-id` != '')
 293                                                                         OR (`item`.`attach` != '') OR `item`.`wall` OR `item`.`origin`)
 294                                                                         AND `item`.`parent` = `thread`.`iid`)
 295                                         ORDER BY `thread`.`iid` LIMIT ?", $days, $last_id, $limit);
 296                         $count = DBA::numRows($r);
 297                         if ($count > 0) {
 298                                 Logger::log("found expired threads: ".$count);
 299                                 while ($thread = DBA::fetch($r)) {
 300                                         $last_id = $thread["iid"];
 301                                         DBA::delete('thread', ['iid' => $thread["iid"]]);
 302                                 }
 303                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 8, $last_id);
 304                         } else {
 305                                 Logger::log("No expired threads found");
 306                         }
 307                         DBA::close($r);
 308                         Logger::log("Done deleting ".$count." expired threads. Last ID: ".$last_id);
 309
 310                         DI::config()->set('system', 'dbclean-last-id-8', $last_id);
 311                 } elseif ($stage == 9) {
 312                         if ($days <= 0) {
 313                                 return;
 314                         }
 315
 316                         $last_id = DI::config()->get('system', 'dbclean-last-id-9', 0);
 317                         $till_id = DI::config()->get('system', 'dbclean-last-id-8', 0);
 318
 319                         Logger::log("Deleting old global item entries from expired threads from ID ".$last_id." to ID ".$till_id);
 320                         $r = DBA::p("SELECT `id` FROM `item` WHERE `uid` = 0 AND
 321                                                 NOT EXISTS (SELECT `guid` FROM `item` AS `i` WHERE `item`.`guid` = `i`.`guid` AND `i`.`uid` != 0) AND
 322                                                 `received` < UTC_TIMESTAMP() - INTERVAL 90 DAY AND `id` >= ? AND `id` <= ?
 323                                         ORDER BY `id` LIMIT ?", $last_id, $till_id, $limit);
 324                         $count = DBA::numRows($r);
 325                         if ($count > 0) {
 326                                 Logger::log("found global item entries from expired threads: ".$count);
 327                                 while ($orphan = DBA::fetch($r)) {
 328                                         $last_id = $orphan["id"];
 329                                         DBA::delete('item', ['id' => $orphan["id"]]);
 330                                 }
 331                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 9, $last_id);
 332                         } else {
 333                                 Logger::log("No global item entries from expired threads");
 334                         }
 335                         DBA::close($r);
 336                         Logger::log("Done deleting ".$count." old global item entries from expired threads. Last ID: ".$last_id);
 337
 338                         DI::config()->set('system', 'dbclean-last-id-9', $last_id);
 339                 } elseif ($stage == 10) {
 340                         $last_id = DI::config()->get('system', 'dbclean-last-id-10', 0);
 341                         $days = intval(DI::config()->get('system', 'dbclean_expire_conversation', 90));
 342
 343                         Logger::log("Deleting old conversations. Last created: ".$last_id);
 344                         $r = DBA::p("SELECT `received`, `item-uri` FROM `conversation`
 345                                         WHERE `received` < UTC_TIMESTAMP() - INTERVAL ? DAY
 346                                         ORDER BY `received` LIMIT ?", $days, $limit);
 347                         $count = DBA::numRows($r);
 348                         if ($count > 0) {
 349                                 Logger::log("found old conversations: ".$count);
 350                                 while ($orphan = DBA::fetch($r)) {
 351                                         $last_id = $orphan["received"];
 352                                         DBA::delete('conversation', ['item-uri' => $orphan["item-uri"]]);
 353                                 }
 354                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 10, $last_id);
 355                         } else {
 356                                 Logger::log("No old conversations found");
 357                         }
 358                         DBA::close($r);
 359                         Logger::log("Done deleting ".$count." conversations. Last created: ".$last_id);
 360
 361                         DI::config()->set('system', 'dbclean-last-id-10', $last_id);
 362                 }
 363         }
 364 }