Max file size for metadata generation

Signed-off-by: Git'Fellow <12234510+solracsf@users.noreply.github.com>

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

Signed-off-by: Git'Fellow <12234510+solracsf@users.noreply.github.com>

From Bytes to MiB

Signed-off-by: Git'Fellow <12234510+solracsf@users.noreply.github.com>

Fix psalm

Signed-off-by: Git'Fellow <12234510+solracsf@users.noreply.github.com>

Rename var to max_filesize

Signed-off-by: Git'Fellow <12234510+solracsf@users.noreply.github.com>
Co-Authored-By: Louis <louis@chmn.me>

Author: solracsf

config/config.sample.php

@@ -1324,6 +1324,15 @@
 	'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

@@ -37,6 +37,9 @@
 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,
@@ -100,6 +103,15 @@ 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) {