feat(PLATFORM-10268): optimize migrateLinksTable.php for pagelinks

maintenance/includes/LoggedUpdateMaintenance.php

@@ -48,9 +48,18 @@ public function execute() {
 			return true;
 		}
 
+		// Fandom-start PLATFORM-10268
+		// log execution time of each migration script, so we could easily find migration bottlenecks
+		$start = microtime( true );
+		$this->output( "Running '$key'\n" );
 		if ( !$this->doDBUpdates() ) {
+			$elapsed = microtime( true ) - $start;
+			$this->output( "'$key' failed after {$elapsed}s\n" );
 			return false;
 		}
+		$elapsed = microtime( true ) - $start;
+		$this->output( "'$key' finished after {$elapsed}s\n" );
+		// Fandom-end
 
 		$db->newInsertQueryBuilder()
 			->insertInto( 'updatelog' )

maintenance/migrateLinksTable.php

@@ -75,6 +75,18 @@ protected function doDBUpdates() {
 		}
 		$highestPageId = $highestPageId[0];
 		$pageId = 0;
+
+		if ( $table === 'pagelinks' ) {
+			$createdLinkTargetRows = $this->fillLinkTargetTable();
+			$this->output( "In total created $createdLinkTargetRows linktarget rows\n" );
+
+			$updatedPageLinksRows = $this->handlePagelinksUpdate();
+			$this->output( "In total updated $updatedPageLinksRows pagelinks rows\n" );
+
+			$updated = $updatedPageLinksRows + $createdLinkTargetRows;
+			$this->output( "Completed normalization of $table, $updated rows updated.\n" );
+			return true;
+		}
 		while ( $pageId <= $highestPageId ) {
 			// Given the indexes and the structure of links tables,
 			// we need to split the update into batches of pages.
@@ -88,6 +100,47 @@ protected function doDBUpdates() {
 		return true;
 	}
 
+	private function fillLinkTargetTable(): int {
+		$batchSize = $this->getBatchSize();
+		$query = "INSERT INTO linktarget(lt_namespace, lt_title)
+					SELECT pl_namespace, pl_title FROM pagelinks
+					WHERE NOT exists (SELECT * FROM linktarget WHERE pl_namespace = lt_namespace AND pl_title = lt_title)
+					GROUP BY pl_namespace, pl_title
+					LIMIT $batchSize";
+		$dbw = $this->getPrimaryDB();
+		$createdRows = 0;
+		while ( true ) {
+			$dbw->query( $query, __METHOD__ );
+			$affectedRows = $dbw->affectedRows();
+			$createdRows += $affectedRows;
+			$this->output( "Created $affectedRows linktarget rows\n" );
+			$this->waitForReplication();
+			if ( $batchSize > $affectedRows ) {
+				return $createdRows;
+			}
+		}
+	}
+
+	private function handlePagelinksUpdate(): int {
+		$batchSize = $this->getBatchSize();
+		$query = "UPDATE pagelinks
+					SET pl_target_id = (SELECT lt_id FROM linktarget WHERE pl_namespace = lt_namespace AND pl_title = lt_title)
+					WHERE pl_target_id IS NULL OR pl_target_id = 0
+					LIMIT $batchSize";
+		$dbw = $this->getPrimaryDB();
+		$updatedRows = 0;
+		while ( true ) {
+			$dbw->query( $query, __METHOD__ );
+			$affectedRows = $dbw->affectedRows();
+			$updatedRows += $affectedRows;
+			$this->output( "Updated $affectedRows pagelinks rows\n" );
+			$this->waitForReplication();
+			if ( $batchSize > $affectedRows ) {
+				return $updatedRows;
+			}
+		}
+	}
+
 	private function handlePageBatch( $lowPageId, $mapping, $table ) {
 		$batchSize = $this->getBatchSize();
 		$targetColumn = $mapping[$table]['target_id'];

maintenance/migrateRevisionCommentTemp.php

@@ -48,8 +48,6 @@ protected function getUpdateKey() {
 	}
 
 	protected function doDBUpdates() {
-		$batchSize = $this->getBatchSize();
-
 		$dbw = $this->getDB( DB_PRIMARY );
 		if ( !$dbw->fieldExists( 'revision', 'rev_comment_id', __METHOD__ ) ) {
 			$this->output( "Run update.php to create rev_comment_id.\n" );
@@ -61,48 +59,36 @@ protected function doDBUpdates() {
 		}
 
 		$this->output( "Merging the revision_comment_temp table into the revision table...\n" );
-		$conds = [];
 		$updated = 0;
-		$sleep = (int)$this->getOption( 'sleep', 0 );
+		$highestRevId = (int)$dbw->newSelectQueryBuilder()
+			->select( 'rev_id' )
+			->from( 'revision' )
+			->limit( 1 )
+			->caller( __METHOD__ )
+			->orderBy( 'rev_id', 'DESC' )
+			->fetchField();
+		$this->output( "Max rev_id $highestRevId.\n" );
+		// Default batchSize from "$this->getBatchSize()" is 200, use 1000 to speed migration up
+		// There is "$this->waitForReplication()" after each batch anyway
+		$batchSize = 1000;
+		$lowId = -1;
+		$highId = $batchSize;
 		while ( true ) {
-			$res = $dbw->newSelectQueryBuilder()
-				->select( [ 'rev_id', 'revcomment_comment_id' ] )
-				->from( 'revision' )
-				->join( 'revision_comment_temp', null, 'rev_id=revcomment_rev' )
-				->where( [ 'rev_comment_id' => 0 ] )
-				->andWhere( $conds )
-				->limit( $batchSize )
-				->orderBy( 'rev_id' )
-				->caller( __METHOD__ )
-				->fetchResultSet();
-
-			$numRows = $res->numRows();
-
-			$last = null;
-			foreach ( $res as $row ) {
-				$last = $row->rev_id;
-				$dbw->newUpdateQueryBuilder()
-					->update( 'revision' )
-					->set( [ 'rev_comment_id' => $row->revcomment_comment_id ] )
-					->where( [ 'rev_id' => $row->rev_id ] )
-					->caller( __METHOD__ )->execute();
-				$updated += $dbw->affectedRows();
-			}
+			$query = "UPDATE revision
+				SET rev_comment_id = (SELECT revcomment_comment_id FROM revision_comment_temp WHERE rev_id=revcomment_rev)
+				WHERE rev_id > $lowId AND rev_id <= $highId";
+			$dbw->query( $query, __METHOD__ );
+			$affected = $dbw->affectedRows();
+			$updated += $affected;
+			$this->output( "Updated $affected revision rows from $lowId to $highId\n" );
+			$this->waitForReplication();
 
-			if ( $numRows < $batchSize ) {
-				// We must have reached the end
+			if ( $highId > $highestRevId ) {
+				// We reached the end
 				break;
 			}
-
-			// @phan-suppress-next-line PhanTypeSuspiciousStringExpression last is not-null when used
-			$this->output( "... rev_id=$last, updated $updated\n" );
-			$conds = [ $dbw->expr( 'rev_id', '>', $last ) ];
-
-			// Sleep between batches for replication to catch up
-			$this->waitForReplication();
-			if ( $sleep > 0 ) {
-				sleep( $sleep );
-			}
+			$lowId = $highId;
+			$highId = $lowId + $batchSize;
 		}
 		$this->output(
 			"Completed merge of revision_comment_temp into the revision table, "