maintenance/fixSlaveDesync.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @ingroup Maintenance
  19  */
  20
  21 require_once( dirname(__FILE__) . '/Maintenance.php' );
  22
  23 class FixSlaveDesync extends Maintenance {
  24         public function __construct() {
  25                 global $wgUseRootUser;
  26                 $wgUseRootUser = true;
  27
  28                 parent::__construct();
  29                 $this->mDescription = "";
  30
  31         }
  32
  33         public function execute() {
  34                 global $slaveIndexes, $wgDBservers;
  35                 $slaveIndexes = array();
  36                 for ( $i = 1; $i < count( $wgDBservers ); $i++ ) {
  37                         if ( wfGetLB()->isNonZeroLoad( $i ) ) {
  38                                 $slaveIndexes[] = $i;
  39                         }
  40                 }
  41
  42                 if ( $this->hasArg() ) {
  43                         $this->desyncFixPage( $this->getArg() );
  44                 } else {
  45                         $dbw = wfGetDB( DB_MASTER );
  46                         $maxPage = $dbw->selectField( 'page', 'MAX(page_id)', false, __METHOD__ );
  47                         $corrupt = $this->findPageLatestCorruption();
  48                         foreach ( $corrupt as $id => $dummy ) {
  49                                 $this->desyncFixPage( $id );
  50                         }
  51                 }
  52         }
  53
  54         /**
  55          * Find all pages that have a corrupted page_latest
  56          * @return array
  57          */
  58         private function findPageLatestCorruption() {
  59                 $desync = array();
  60                 $n = 0;
  61                 $dbw = wfGetDB( DB_MASTER );
  62                 $masterIDs = array();
  63                 $res = $dbw->select( 'page', array( 'page_id', 'page_latest' ), array( 'page_id<6054123' ), __METHOD__ );
  64                 $this->output( "Number of pages: " . $dbw->numRows( $res ) . "\n" );
  65                 foreach ( $res as $row ) {
  66                         $masterIDs[$row->page_id] = $row->page_latest;
  67                         if ( !( ++$n % 10000 ) ) {
  68                                 $this->output( "$n\r" );
  69                         }
  70                 }
  71                 $this->output( "\n" );
  72                 $dbw->freeResult( $res );
  73
  74                 global $slaveIndexes;
  75                 foreach ( $slaveIndexes as $i ) {
  76                         $db = wfGetDB( $i );
  77                         $res = $db->select( 'page', array( 'page_id', 'page_latest' ), array( 'page_id<6054123' ), __METHOD__ );
  78                         foreach ( $res as $row ) {
  79                                 if ( isset( $masterIDs[$row->page_id] ) && $masterIDs[$row->page_id] != $row->page_latest ) {
  80                                         $desync[$row->page_id] = true;
  81                                         $this->output( $row->page_id . "\t" );
  82                                 }
  83                         }
  84                         $db->freeResult( $res );
  85                 }
  86                 $this->output( "\n" );
  87                 return $desync;
  88         }
  89
  90         /**
  91          * Fix a broken page entry
  92          * @param $pageID int The page_id to fix
  93          */
  94         private function desyncFixPage( $pageID ) {
  95                 global $slaveIndexes;
  96
  97                 # Check for a corrupted page_latest
  98                 $dbw = wfGetDB( DB_MASTER );
  99                 $dbw->begin();
 100                 $realLatest = $dbw->selectField( 'page', 'page_latest', array( 'page_id' => $pageID ),
 101                         __METHOD__, 'FOR UPDATE' );
 102                 #list( $masterFile, $masterPos ) = $dbw->getMasterPos();
 103                 $found = false;
 104                 foreach ( $slaveIndexes as $i ) {
 105                         $db = wfGetDB( $i );
 106                         /*
 107                         if ( !$db->masterPosWait( $masterFile, $masterPos, 10 ) ) {
 108                                    $this->output( "Slave is too lagged, aborting\n" );
 109                                    $dbw->commit();
 110                                    sleep(10);
 111                                    return;
 112                         }*/
 113                         $latest = $db->selectField( 'page', 'page_latest', array( 'page_id' => $pageID ), __METHOD__ );
 114                         $max = $db->selectField( 'revision', 'MAX(rev_id)', false, __METHOD__ );
 115                         if ( $latest != $realLatest && $realLatest < $max ) {
 116                                 $this->output( "page_latest corrupted in page $pageID, server $i\n" );
 117                                 $found = true;
 118                                 break;
 119                         }
 120                 }
 121                 if ( !$found ) {
 122                         $this->output( "page_id $pageID seems fine\n" );
 123                         $dbw->commit();
 124                         return;
 125                 }
 126
 127                 # Find the missing revisions
 128                 $res = $dbw->select( 'revision', array( 'rev_id' ), array( 'rev_page' => $pageID ),
 129                         __METHOD__, 'FOR UPDATE' );
 130                 $masterIDs = array();
 131                 foreach ( $res as $row ) {
 132                         $masterIDs[] = $row->rev_id;
 133                 }
 134                 $dbw->freeResult( $res );
 135
 136                 $res = $db->select( 'revision', array( 'rev_id' ), array( 'rev_page' => $pageID ), __METHOD__ );
 137                 $slaveIDs = array();
 138                 foreach ( $res as $row ) {
 139                         $slaveIDs[] = $row->rev_id;
 140                 }
 141                 $db->freeResult( $res );
 142                 if ( count( $masterIDs ) < count( $slaveIDs ) ) {
 143                         $missingIDs = array_diff( $slaveIDs, $masterIDs );
 144                         if ( count( $missingIDs ) ) {
 145                                 $this->output( "Found " . count( $missingIDs ) . " lost in master, copying from slave... " );
 146                                 $dbFrom = $db;
 147                                 $found = true;
 148                                 $toMaster = true;
 149                         } else {
 150                                 $found = false;
 151                         }
 152                 } else {
 153                         $missingIDs = array_diff( $masterIDs, $slaveIDs );
 154                         if ( count( $missingIDs ) ) {
 155                                 $this->output( "Found " . count( $missingIDs ) . " missing revision(s), copying from master... " );
 156                                 $dbFrom = $dbw;
 157                                 $found = true;
 158                                 $toMaster = false;
 159                         } else {
 160                                 $found = false;
 161                         }
 162                 }
 163
 164                 if ( $found ) {
 165                         foreach ( $missingIDs as $rid ) {
 166                                 $this->output( "$rid " );
 167                                 # Revision
 168                                 $row = $dbFrom->selectRow( 'revision', '*', array( 'rev_id' => $rid ), __METHOD__ );
 169                                 if ( $toMaster ) {
 170                                         $id = $dbw->selectField( 'revision', 'rev_id', array( 'rev_id' => $rid ),
 171                                                 __METHOD__, 'FOR UPDATE' );
 172                                         if ( $id ) {
 173                                                 $this->output( "Revision already exists\n" );
 174                                                 $found = false;
 175                                                 break;
 176                                         } else {
 177                                                 $dbw->insert( 'revision', get_object_vars( $row ), __METHOD__, 'IGNORE' );
 178                                         }
 179                                 } else {
 180                                         foreach ( $slaveIndexes as $i ) {
 181                                                 $db = wfGetDB( $i );
 182                                                 $db->insert( 'revision', get_object_vars( $row ), __METHOD__, 'IGNORE' );
 183                                         }
 184                                 }
 185
 186                                 # Text
 187                                 $row = $dbFrom->selectRow( 'text', '*', array( 'old_id' => $row->rev_text_id ), __METHOD__ );
 188                                 if ( $toMaster ) {
 189                                         $dbw->insert( 'text', get_object_vars( $row ), __METHOD__, 'IGNORE' );
 190                                 } else {
 191                                         foreach ( $slaveIndexes as $i ) {
 192                                                 $db = wfGetDB( $i );
 193                                                 $db->insert( 'text', get_object_vars( $row ), __METHOD__, 'IGNORE' );
 194                                         }
 195                                 }
 196                         }
 197                         $this->output( "done\n" );
 198                 }
 199
 200                 if ( $found ) {
 201                         $this->output( "Fixing page_latest... " );
 202                         if ( $toMaster ) {
 203                                 #$dbw->update( 'page', array( 'page_latest' => $realLatest ), array( 'page_id' => $pageID ), __METHOD__ );
 204                         } else {
 205                                 foreach ( $slaveIndexes as $i ) {
 206                                         $db = wfGetDB( $i );
 207                                         $db->update( 'page', array( 'page_latest' => $realLatest ), array( 'page_id' => $pageID ), __METHOD__ );
 208                                 }
 209                         }
 210                         $this->output( "done\n" );
 211                 }
 212                 $dbw->commit();
 213         }
 214 }
 215
 216 $maintClass = "FixSlaveDesync";
 217 require_once( DO_MAINTENANCE );