Transparent* database sharding

.github/workflows/phpunit-mysql-sharding.yml

@@ -0,0 +1,184 @@
+# This workflow is provided via the organization template repository
+#
+# https://github.com/nextcloud/.github
+# https://docs.github.com/en/actions/learn-github-actions/sharing-workflows-with-your-organization
+#
+# SPDX-FileCopyrightText: 2022-2024 Nextcloud GmbH and Nextcloud contributors
+# SPDX-License-Identifier: MIT
+
+name: PHPUnit sharding
+
+on:
+  pull_request:
+  schedule:
+    - cron: "5 2 * * *"
+
+permissions:
+  contents: read
+
+concurrency:
+  group: phpunit-mysql-sharding-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  changes:
+    runs-on: ubuntu-latest-low
+
+    outputs:
+      src: ${{ steps.changes.outputs.src }}
+
+    steps:
+      - uses: dorny/paths-filter@de90cc6fb38fc0963ad72b210f1f284cd68cea36 # v3.0.2
+        id: changes
+        continue-on-error: true
+        with:
+          filters: |
+            src:
+              - '.github/workflows/**'
+              - '3rdparty/**'
+              - '**/appinfo/**'
+              - '**/lib/**'
+              - '**/templates/**'
+              - '**/tests/**'
+              - 'vendor/**'
+              - 'vendor-bin/**'
+              - '.php-cs-fixer.dist.php'
+              - 'composer.json'
+              - 'composer.lock'
+              - '**.php'
+
+  phpunit-mysql:
+    runs-on: ubuntu-latest
+
+    needs: changes
+    if: needs.changes.outputs.src != 'false'
+
+    strategy:
+      matrix:
+        php-versions: ['8.1']
+        mysql-versions: ['8.4']
+
+    name: Sharding - MySQL ${{ matrix.mysql-versions }} (PHP ${{ matrix.php-versions }}) - database tests
+
+    services:
+      cache:
+        image: ghcr.io/nextcloud/continuous-integration-redis:latest
+        ports:
+          - 6379:6379/tcp
+        options: --health-cmd="redis-cli ping" --health-interval=10s --health-timeout=5s --health-retries=3
+
+      mysql:
+        image: ghcr.io/nextcloud/continuous-integration-mysql-${{ matrix.mysql-versions }}:latest
+        ports:
+          - 4444:3306/tcp
+        env:
+          MYSQL_ROOT_PASSWORD: rootpassword
+          MYSQL_USER: oc_autotest
+          MYSQL_PASSWORD: nextcloud
+          MYSQL_DATABASE: oc_autotest
+        options: --health-cmd="mysqladmin ping" --health-interval 5s --health-timeout 2s --health-retries 10
+      shard1:
+        image: ghcr.io/nextcloud/continuous-integration-mysql-${{ matrix.mysql-versions }}:latest
+        ports:
+          - 5001:3306/tcp
+        env:
+          MYSQL_ROOT_PASSWORD: rootpassword
+          MYSQL_USER: oc_autotest
+          MYSQL_PASSWORD: nextcloud
+          MYSQL_DATABASE: nextcloud
+        options: --health-cmd="mysqladmin ping" --health-interval 5s --health-timeout 2s --health-retries 10
+      shard2:
+        image: ghcr.io/nextcloud/continuous-integration-mysql-${{ matrix.mysql-versions }}:latest
+        ports:
+          - 5002:3306/tcp
+        env:
+          MYSQL_ROOT_PASSWORD: rootpassword
+          MYSQL_USER: oc_autotest
+          MYSQL_PASSWORD: nextcloud
+          MYSQL_DATABASE: nextcloud
+        options: --health-cmd="mysqladmin ping" --health-interval 5s --health-timeout 2s --health-retries 10
+      shard3:
+        image: ghcr.io/nextcloud/continuous-integration-mysql-${{ matrix.mysql-versions }}:latest
+        ports:
+          - 5003:3306/tcp
+        env:
+          MYSQL_ROOT_PASSWORD: rootpassword
+          MYSQL_USER: oc_autotest
+          MYSQL_PASSWORD: nextcloud
+          MYSQL_DATABASE: nextcloud
+        options: --health-cmd="mysqladmin ping" --health-interval 5s --health-timeout 2s --health-retries 10
+      shard4:
+        image: ghcr.io/nextcloud/continuous-integration-mysql-${{ matrix.mysql-versions }}:latest
+        ports:
+          - 5004:3306/tcp
+        env:
+          MYSQL_ROOT_PASSWORD: rootpassword
+          MYSQL_USER: oc_autotest
+          MYSQL_PASSWORD: nextcloud
+          MYSQL_DATABASE: nextcloud
+        options: --health-cmd="mysqladmin ping" --health-interval 5s --health-timeout 2s --health-retries 10
+
+    steps:
+      - name: Checkout server
+        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332
+        with:
+          submodules: true
+
+      - name: Set up php ${{ matrix.php-versions }}
+        uses: shivammathur/setup-php@2e947f1f6932d141d076ca441d0e1e881775e95b #v2.31.0
+        with:
+          php-version: ${{ matrix.php-versions }}
+          # https://docs.nextcloud.com/server/stable/admin_manual/installation/source_installation.html#prerequisites-for-manual-installation
+          extensions: bz2, ctype, curl, dom, fileinfo, gd, iconv, intl, json, libxml, mbstring, openssl, pcntl, posix, redis, session, simplexml, xmlreader, xmlwriter, zip, zlib, mysql, pdo_mysql
+          coverage: ${{ matrix.coverage && 'xdebug' || 'none' }}
+          ini-file: development
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Set up dependencies
+        run: composer i
+
+      - name: Enable ONLY_FULL_GROUP_BY MySQL option
+        run: |
+          echo "SET GLOBAL sql_mode=(SELECT CONCAT(@@sql_mode,',ONLY_FULL_GROUP_BY'));" | mysql -h 127.0.0.1 -P 4444 -u root -prootpassword
+          echo "SELECT @@sql_mode;" | mysql -h 127.0.0.1 -P 4444 -u root -prootpassword
+
+      - name: Set up Nextcloud
+        env:
+          DB_PORT: 4444
+          SHARDING: 1
+        run: |
+          mkdir data
+          cp tests/redis.config.php config/
+          cp tests/preseed-config.php config/config.php
+          ./occ maintenance:install --verbose --database=mysql --database-name=nextcloud --database-host=127.0.0.1 --database-port=$DB_PORT --database-user=root --database-pass=rootpassword --admin-user admin --admin-pass admin
+          php -f tests/enable_all.php | grep -i -C9999 error && echo "Error during app setup" && exit 1 || exit 0
+
+      - name: PHPUnit
+        run: composer run test:db ${{ matrix.coverage && ' -- --coverage-clover ./clover.db.xml' || '' }}
+
+      - name: Upload db code coverage
+        if: ${{ !cancelled() && matrix.coverage }}
+        uses: codecov/codecov-action@v4.1.1
+        with:
+          files: ./clover.db.xml
+          flags: phpunit-mysql
+
+      - name: Print logs
+        if: always()
+        run: |
+          cat data/nextcloud.log
+
+  summary:
+    permissions:
+      contents: none
+    runs-on: ubuntu-latest-low
+    needs: [changes, phpunit-mysql]
+
+    if: always()
+
+    name: phpunit-mysql-summary
+
+    steps:
+      - name: Summary status
+        run: if ${{ needs.changes.outputs.src != 'false' && needs.phpunit-mysql.result != 'success' }}; then exit 1; fi

apps/dav/lib/DAV/CustomPropertiesBackend.php

@@ -364,16 +364,16 @@ private function getPublishedProperties(string $path, array $requestedProperties
 	private function cacheDirectory(string $path, Directory $node): void {
 		$prefix = ltrim($path . '/', '/');
 		$query = $this->connection->getQueryBuilder();
-		$query->select('name', 'propertypath', 'propertyname', 'propertyvalue', 'valuetype')
+		$query->select('name', 'p.propertypath', 'p.propertyname', 'p.propertyvalue', 'p.valuetype')
 			->from('filecache', 'f')
-			->leftJoin('f', 'properties', 'p', $query->expr()->andX(
-				$query->expr()->eq('propertypath', $query->func()->concat(
-					$query->createNamedParameter($prefix),
-					'name'
-				)),
-				$query->expr()->eq('userid', $query->createNamedParameter($this->user->getUID()))
-			))
-			->where($query->expr()->eq('parent', $query->createNamedParameter($node->getInternalFileId(), IQueryBuilder::PARAM_INT)));
+			->hintShardKey('storage', $node->getNode()->getMountPoint()->getNumericStorageId())
+			->leftJoin('f', 'properties', 'p', $query->expr()->eq('p.propertypath', $query->func()->concat(
+				$query->createNamedParameter($prefix),
+				'f.name'
+			)),
+			)
+			->where($query->expr()->eq('parent', $query->createNamedParameter($node->getInternalFileId(), IQueryBuilder::PARAM_INT)))
+			->andWhere($query->expr()->eq('p.userid', $query->createNamedParameter($this->user->getUID())));
 		$result = $query->executeQuery();
 
 		$propsByPath = [];

apps/files/lib/BackgroundJob/DeleteOrphanedItems.php

@@ -52,34 +52,86 @@ public function run($argument) {
 	 * @param string $typeCol
 	 * @return int Number of deleted entries
 	 */
-	protected function cleanUp($table, $idCol, $typeCol) {
+	protected function cleanUp(string $table, string $idCol, string $typeCol): int {
 		$deletedEntries = 0;
 
-		$query = $this->connection->getQueryBuilder();
-		$query->select('t1.' . $idCol)
-			->from($table, 't1')
-			->where($query->expr()->eq($typeCol, $query->expr()->literal('files')))
-			->andWhere($query->expr()->isNull('t2.fileid'))
-			->leftJoin('t1', 'filecache', 't2', $query->expr()->eq($query->expr()->castColumn('t1.' . $idCol, IQueryBuilder::PARAM_INT), 't2.fileid'))
-			->groupBy('t1.' . $idCol)
-			->setMaxResults(self::CHUNK_SIZE);
-
 		$deleteQuery = $this->connection->getQueryBuilder();
 		$deleteQuery->delete($table)
-			->where($deleteQuery->expr()->in($idCol, $deleteQuery->createParameter('objectid')));
+			->where($deleteQuery->expr()->eq($idCol, $deleteQuery->createParameter('objectid')));
+
+		if ($this->connection->getShardDefinition('filecache')) {
+			$sourceIdChunks = $this->getItemIds($table, $idCol, $typeCol, 1000);
+			foreach ($sourceIdChunks as $sourceIdChunk) {
+				$deletedSources = $this->findMissingSources($sourceIdChunk);
+				$deleteQuery->setParameter('objectid', $deletedSources, IQueryBuilder::PARAM_INT_ARRAY);
+				$deletedEntries += $deleteQuery->executeStatement();
+			}
+		} else {
+			$query = $this->connection->getQueryBuilder();
+			$query->select('t1.' . $idCol)
+				->from($table, 't1')
+				->where($query->expr()->eq($typeCol, $query->expr()->literal('files')))
+				->leftJoin('t1', 'filecache', 't2', $query->expr()->eq($query->expr()->castColumn('t1.' . $idCol, IQueryBuilder::PARAM_INT), 't2.fileid'))
+				->andWhere($query->expr()->isNull('t2.fileid'))
+				->groupBy('t1.' . $idCol)
+				->setMaxResults(self::CHUNK_SIZE);
+
+			$deleteQuery = $this->connection->getQueryBuilder();
+			$deleteQuery->delete($table)
+				->where($deleteQuery->expr()->in($idCol, $deleteQuery->createParameter('objectid')));
 
-		$deletedInLastChunk = self::CHUNK_SIZE;
-		while ($deletedInLastChunk === self::CHUNK_SIZE) {
-			$chunk = $query->executeQuery()->fetchAll(\PDO::FETCH_COLUMN);
-			$deletedInLastChunk = count($chunk);
+			$deletedInLastChunk = self::CHUNK_SIZE;
+			while ($deletedInLastChunk === self::CHUNK_SIZE) {
+				$chunk = $query->executeQuery()->fetchAll(\PDO::FETCH_COLUMN);
+				$deletedInLastChunk = count($chunk);
 
-			$deleteQuery->setParameter('objectid', $chunk, IQueryBuilder::PARAM_INT_ARRAY);
-			$deletedEntries += $deleteQuery->executeStatement();
+				$deleteQuery->setParameter('objectid', $chunk, IQueryBuilder::PARAM_INT_ARRAY);
+				$deletedEntries += $deleteQuery->executeStatement();
+			}
 		}
 
 		return $deletedEntries;
 	}
 
+	/**
+	 * @param string $table
+	 * @param string $idCol
+	 * @param string $typeCol
+	 * @param int $chunkSize
+	 * @return \Iterator<int[]>
+	 * @throws \OCP\DB\Exception
+	 */
+	private function getItemIds(string $table, string $idCol, string $typeCol, int $chunkSize): \Iterator {
+		$query = $this->connection->getQueryBuilder();
+		$query->select($idCol)
+			->from($table)
+			->where($query->expr()->eq($typeCol, $query->expr()->literal('files')))
+			->groupBy($idCol)
+			->andWhere($query->expr()->gt($idCol, $query->createParameter('min_id')))
+			->setMaxResults($chunkSize);
+
+		$minId = 0;
+		while (true) {
+			$query->setParameter('min_id', $minId);
+			$rows = $query->executeQuery()->fetchAll(\PDO::FETCH_COLUMN);
+			if (count($rows) > 0) {
+				$minId = $rows[count($rows) - 1];
+				yield $rows;
+			} else {
+				break;
+			}
+		}
+	}
+
+	private function findMissingSources(array $ids): array {
+		$qb = $this->connection->getQueryBuilder();
+		$qb->select('fileid')
+			->from('filecache')
+			->where($qb->expr()->in('fileid', $qb->createNamedParameter($ids, IQueryBuilder::PARAM_INT_ARRAY)));
+		$found = $qb->executeQuery()->fetchAll(\PDO::FETCH_COLUMN);
+		return array_diff($ids, $found);
+	}
+
 	/**
 	 * Deleting orphaned system tag mappings
 	 *