Merge pull request #42800 from nextcloud/metaGenMemLimit

enh(metadata): Introduce a memory limit for metadata generation

config/config.sample.php

@@ -1373,6 +1373,15 @@
 	'OC\Preview\XBitmap',
 ],
 
+/**
+ * Maximum file size for metadata generation.
+ * If a file exceeds this size, metadata generation will be skipped.
+ * Note: memory equivalent to this size will be used for metadata generation.
+ *
+ * Default: 256 megabytes.
+ */
+'metadata_max_filesize' => 256,
+
 /**
  * LDAP
  *

core/BackgroundJobs/GenerateMetadataJob.php

@@ -16,12 +16,17 @@
 use OCP\FilesMetadata\Exceptions\FilesMetadataNotFoundException;
 use OCP\FilesMetadata\IFilesMetadataManager;
 use OCP\IAppConfig;
+use OCP\IConfig;
 use OCP\IUserManager;
 use Psr\Log\LoggerInterface;
 
 class GenerateMetadataJob extends TimedJob {
+	// Default file size limit for metadata generation (MBytes).
+	protected const DEFAULT_MAX_FILESIZE = 256;
+
 	public function __construct(
 		ITimeFactory $time,
+		private IConfig $config,
 		private IAppConfig $appConfig,
 		private IRootFolder $rootFolder,
 		private IUserManager $userManager,
@@ -88,6 +93,15 @@ private function scanFolder(Folder $folder): void {
 				continue;
 			}
 
+			// Don't generate metadata for files bigger than configured metadata_max_filesize
+			// Files are loaded in memory so very big files can lead to an OOM on the server
+			$nodeSize = $node->getSize();
+			$nodeLimit = $this->config->getSystemValueInt('metadata_max_filesize', self::DEFAULT_MAX_FILESIZE);
+			if ($nodeSize > $nodeLimit * 1000000) {
+				$this->logger->debug("Skipping generating metadata for fileid " . $node->getId() . " as its size exceeds configured 'metadata_max_filesize'.");
+				continue;
+			}
+
 			try {
 				$this->filesMetadataManager->getMetadata($node->getId(), false);
 			} catch (FilesMetadataNotFoundException) {