Merge pull request #7: Release 0.9.3-beta.4

Release 0.9.3-beta.4

CHANGELOG.md

@@ -7,6 +7,19 @@ the app store accepts a `-beta` suffix, but Nextcloud servers then only offer th
 beta/daily update channel, which would hide the app from most admins. The stability stage is
 instead carried by the app description, the git tag, the release, and this file.
 
+## [0.9.3-beta.4] - 2026-10-08
+
+Bugfix release.
+
+### Fixed
+- A backup failed with "cannot execute INSERT in a read-only transaction" on PostgreSQL as soon
+  as the database dump filled a whole pack (32 MiB): the pack's index cache row was written
+  inside the read-only dump transaction. On other databases that row was rolled back silently.
+- Backups to a remote location slowed down as the repository grew: every backup step listed the
+  repository's index with a round trip per file. Over SFTP to a Hetzner Storage Box a 60 GiB
+  repository already lost a large part of each step to it; backing up continuously
+  (`occ backup:run`) went from 45 KiB/s to 5.8 MB/s with the fix.
+
 ## [0.9.2-beta.3] - 2026-10-08
 
 Bugfix release.
@@ -122,6 +135,7 @@ and not implemented yet.
 - See the "Storage backend experiences" section of the README for provider-specific quirks found
   so far (e.g. Hetzner Storage Box sub-accounts).
 
+[0.9.3-beta.4]: https://github.com/dillardblom/ng_backup/releases/tag/v0.9.3-beta.4
 [0.9.2-beta.3]: https://github.com/dillardblom/ng_backup/releases/tag/v0.9.2-beta.3
 [0.9.1-beta.2]: https://github.com/dillardblom/ng_backup/releases/tag/v0.9.1-beta.2
 [0.9.0-beta.1]: https://github.com/dillardblom/ng_backup/releases/tag/v0.9.0-beta.1

README.md

@@ -7,7 +7,7 @@ installations and hosted Nextcloud. (Choosing SMB as a storage backend is the on
 see "Storage backend experiences" below — that requirement comes from files_external, not from
 ng_backup, and applies to any Nextcloud app that writes to an SMB share.)
 
-**Status: beta (0.9.2).** Tested thoroughly, usable alongside your current backup, but not yet a
+**Status: beta (0.9.3).** Tested thoroughly, usable alongside your current backup, but not yet a
 drop-in replacement for it. See "Known limitations" below, `PLAN.md` for the design and
 `docs/HOWTO.md` for a step-by-step guide from setup to a tested restore.
 

appinfo/info.xml

@@ -9,7 +9,7 @@ directory to S3, SFTP, WebDAV, FTP, SMB or a local path, with restore of files a
 single users (via user_migration) and of the whole server onto a fresh installation. Tested
 thoroughly and usable alongside your current backup, but not yet a drop-in replacement for it.
 See the README for known limitations and the recovery kit before relying on it.]]></description>
-    <version>0.9.2</version>
+    <version>0.9.3</version>
     <licence>agpl</licence>
     <author>Dillard Blom</author>
     <namespace>NgBackup</namespace>

lib/Backend/IBackend.php

@@ -37,5 +37,15 @@ interface IBackend {
 	/** @return list<string> object paths below $prefix (recursive) */
 	public function list(string $prefix): array;
 
+	/**
+	 * Object paths directly in $dir, for a folder that only holds objects (index/, snapshots/,
+	 * catalog/, trash/info/). One listing, without checking the type of every entry: on a remote
+	 * storage list() costs a round trip per entry, which for index/ (one file per pack) grows
+	 * with the size of the repository.
+	 *
+	 * @return list<string>
+	 */
+	public function listFiles(string $dir): array;
+
 	public function delete(string $path): void;
 }

lib/Backend/LocalBackend.php

@@ -107,6 +107,25 @@ final class LocalBackend implements IBackend {
 		return $result;
 	}
 
+	public function listFiles(string $dir): array {
+		$base = $this->abs($dir);
+		if (!is_dir($base)) {
+			return [];
+		}
+		$result = [];
+		$names = scandir($base);
+		if ($names === false) {
+			throw new BackendException('Cannot list ' . $dir);
+		}
+		foreach ($names as $name) {
+			if ($name !== '.' && $name !== '..' && !StorageBackend::isTemporary($name)) {
+				$result[] = trim($dir, '/') . '/' . $name;
+			}
+		}
+		sort($result);
+		return $result;
+	}
+
 	public function delete(string $path): void {
 		@unlink($this->abs($path));
 	}

lib/Backend/StorageBackend.php

@@ -133,6 +133,28 @@ final class StorageBackend implements IBackend {
 		}
 	}
 
+	public function listFiles(string $dir): array {
+		$abs = $this->abs($dir);
+		if (!$this->storage->is_dir($abs)) {
+			return []; // not created yet
+		}
+		$dh = $this->storage->opendir($abs);
+		if ($dh === false) {
+			// An existing folder read as empty would look like a new repository.
+			throw new BackendException('Cannot list ' . $dir);
+		}
+		$cut = $this->base === '' ? 0 : strlen($this->base) + 1;
+		$result = [];
+		while (($name = readdir($dh)) !== false) {
+			if ($name !== '.' && $name !== '..' && !self::isTemporary($name)) {
+				$result[] = substr($abs . '/' . $name, $cut);
+			}
+		}
+		closedir($dh);
+		sort($result);
+		return $result;
+	}
+
 	private function listInto(string $dir, array &$result): void {
 		if (!$this->storage->is_dir($dir)) {
 			return;

lib/Db/DbDumper.php

@@ -72,50 +72,57 @@ final class DbDumper {
 	public function dumpStep(Repository $repo, array $state, float $deadline, array &$stats): array {
 		$state['steps']++;
 		$sm = $this->db->createSchemaManager();
-		$this->db->setTransactionIsolation(TransactionIsolationLevel::REPEATABLE_READ);
-		$this->db->beginTransaction();
+		// A pack that fills up during the dump is uploaded right away; its index cache row must
+		// wait until the read-only transaction is over.
+		$repo->holdIndexCache();
 		try {
-			if ($this->db->getDatabasePlatform() instanceof \Doctrine\DBAL\Platforms\PostgreSQLPlatform) {
-				$this->db->executeStatement('SET TRANSACTION READ ONLY');
-			}
-			while ($state['i'] < count($state['tables'])) {
-				$table = $state['tables'][$state['i']];
-				$schema = $sm->introspectTable($table);
-				$writer = $repo->blobWriter($stats);
-				$hash = $state['hash'] !== null ? unserialize(base64_decode($state['hash']), ['allowed_classes' => [\HashContext::class]]) : hash_init('sha256');
-				if ($state['after'] === null && $state['segments'] === []) {
-					$writer->write($this->header($schema) . "\n");
+			$this->db->setTransactionIsolation(TransactionIsolationLevel::REPEATABLE_READ);
+			$this->db->beginTransaction();
+			try {
+				if ($this->db->getDatabasePlatform() instanceof \Doctrine\DBAL\Platforms\PostgreSQLPlatform) {
+					$this->db->executeStatement('SET TRANSACTION READ ONLY');
 				}
-				$complete = true;
-				foreach ($this->rowLines($schema, $state['after']) as [$key, $line]) {
-					$writer->write($line . "\n");
-					hash_update($hash, $line . "\n");
-					$state['rows']++;
-					$state['after'] = $key;
-					if (microtime(true) >= $deadline) {
-						$complete = false;
+				while ($state['i'] < count($state['tables'])) {
+					$table = $state['tables'][$state['i']];
+					$schema = $sm->introspectTable($table);
+					$writer = $repo->blobWriter($stats);
+					$hash = $state['hash'] !== null ? unserialize(base64_decode($state['hash']), ['allowed_classes' => [\HashContext::class]]) : hash_init('sha256');
+					if ($state['after'] === null && $state['segments'] === []) {
+						$writer->write($this->header($schema) . "\n");
+					}
+					$complete = true;
+					foreach ($this->rowLines($schema, $state['after']) as [$key, $line]) {
+						$writer->write($line . "\n");
+						hash_update($hash, $line . "\n");
+						$state['rows']++;
+						$state['after'] = $key;
+						if (microtime(true) >= $deadline) {
+							$complete = false;
+							break;
+						}
+					}
+					$state['segments'][] = $writer->finish();
+					$state['bytes'] += $writer->bytes();
+					if (!$complete) {
+						$state['hash'] = base64_encode(serialize($hash));
+						break;
+					}
+					$state['result'][$table] = ['blobs' => array_merge(...$state['segments']), 'rows' => $state['rows'],
+						'bytes' => $state['bytes'], 'sha256' => hash_final($hash)];
+					$state = array_merge($state, ['i' => $state['i'] + 1, 'after' => null, 'segments' => [], 'rows' => 0, 'bytes' => 0, 'hash' => null]);
+					if (microtime(true) >= $deadline && $state['i'] < count($state['tables'])) {
 						break;
 					}
 				}
-				$state['segments'][] = $writer->finish();
-				$state['bytes'] += $writer->bytes();
-				if (!$complete) {
-					$state['hash'] = base64_encode(serialize($hash));
-					break;
+				if ($state['i'] >= count($state['tables'])) {
+					$state['sequences'] = $this->sequenceStates($state['tables']);
+					$state['done'] = true;
 				}
-				$state['result'][$table] = ['blobs' => array_merge(...$state['segments']), 'rows' => $state['rows'],
-					'bytes' => $state['bytes'], 'sha256' => hash_final($hash)];
-				$state = array_merge($state, ['i' => $state['i'] + 1, 'after' => null, 'segments' => [], 'rows' => 0, 'bytes' => 0, 'hash' => null]);
-				if (microtime(true) >= $deadline && $state['i'] < count($state['tables'])) {
-					break;
-				}
-			}
-			if ($state['i'] >= count($state['tables'])) {
-				$state['sequences'] = $this->sequenceStates($state['tables']);
-				$state['done'] = true;
+			} finally {
+				$this->db->rollBack();
 			}
 		} finally {
-			$this->db->rollBack();
+			$repo->releaseIndexCache();
 		}
 		$repo->flushPacks();
 		return $state;

lib/Repository/BlobIndex.php

@@ -25,6 +25,8 @@ final class BlobIndex {
 	private array $pending = [];
 
 	private ?IIndexCache $cache = null;
+	/** @var list<\Closure():void>|null cache writes held back by holdCache(), or null when writing directly */
+	private ?array $heldCacheWrites = null;
 	private string $repositoryId = '';
 	/** @var int index files downloaded by the last load() (for tests and diagnostics) */
 	public int $downloaded = 0;
@@ -40,6 +42,46 @@ final class BlobIndex {
 		$this->repositoryId = $repositoryId;
 	}
 
+	/**
+	 * Hold back cache writes until releaseCache(). The database dump reads inside a read-only
+	 * transaction on the same connection that the cache writes to: a write there fails on
+	 * PostgreSQL and would be rolled back with the dump transaction elsewhere.
+	 */
+	public function holdCache(): void {
+		$this->heldCacheWrites ??= [];
+	}
+
+	/** Write the cache changes held back since holdCache(). */
+	public function releaseCache(): void {
+		$writes = $this->heldCacheWrites ?? [];
+		$this->heldCacheWrites = null;
+		foreach ($writes as $write) {
+			$write();
+		}
+	}
+
+	/** @param list<array{id:string, offset:int, length:int, raw:int, flags:int}> $entries */
+	private function cachePut(string $packId, array $entries): void {
+		if ($this->heldCacheWrites !== null) {
+			$this->heldCacheWrites[] = function () use ($packId, $entries): void {
+				$this->cache?->put($this->repositoryId, $packId, $entries);
+			};
+			return;
+		}
+		$this->cache?->put($this->repositoryId, $packId, $entries);
+	}
+
+	/** @param list<string> $packIds */
+	private function cacheRemove(array $packIds): void {
+		if ($this->heldCacheWrites !== null) {
+			$this->heldCacheWrites[] = function () use ($packIds): void {
+				$this->cache?->remove($this->repositoryId, $packIds);
+			};
+			return;
+		}
+		$this->cache?->remove($this->repositoryId, $packIds);
+	}
+
 	/**
 	 * Load the index: one listing of index/, entries from the local cache where available,
 	 * and only the index files that are not cached yet are downloaded. Cached packs that no
@@ -50,7 +92,7 @@ final class BlobIndex {
 		$cached = $verify ? [] : ($this->cache?->all($this->repositoryId) ?? []);
 		$present = [];
 		$this->downloaded = 0;
-		foreach ($this->backend->list('index') as $path) {
+		foreach ($this->backend->listFiles('index') as $path) {
 			$packId = basename($path);
 			$present[$packId] = true;
 			if (isset($cached[$packId])) {
@@ -112,7 +154,7 @@ final class BlobIndex {
 		fwrite($stream, $this->cipher->encryptString(json_encode($entries, JSON_THROW_ON_ERROR), 'index:' . $packId));
 		rewind($stream);
 		$this->backend->put('index/' . $packId, $stream);
-		$this->cache?->put($this->repositoryId, $packId, $entries);
+		$this->cachePut($packId, $entries);
 		foreach ($entries as $e) {
 			$this->blobs[$e['id']] = [$packId, $e['offset'], $e['length'], $e['raw'], $e['flags']];
 			unset($this->pending[$e['id']]);
@@ -131,7 +173,7 @@ final class BlobIndex {
 	/** Forget a pack: delete its index file and cache row. Blobs that now live elsewhere are kept. */
 	public function removePack(string $packId): void {
 		$this->backend->delete('index/' . $packId);
-		$this->cache?->remove($this->repositoryId, [$packId]);
+		$this->cacheRemove([$packId]);
 		foreach ($this->blobs as $id => $loc) {
 			if ($loc[0] === $packId) {
 				unset($this->blobs[$id]);

lib/Repository/Catalog.php

@@ -128,7 +128,7 @@ final class Catalog {
 
 	/** @return list<int> */
 	private function generations(): array {
-		return array_map(fn ($p) => (int)basename($p), $this->backend->list('catalog'));
+		return array_map(fn ($p) => (int)basename($p), $this->backend->listFiles('catalog'));
 	}
 
 	private function path(int $gen): string {

lib/Repository/Repository.php

@@ -92,8 +92,8 @@ final class Repository {
 	private function loadCatalog(?ICatalogAnchor $anchor): void {
 		$this->catalog = new Catalog($this->backend, $this->cipher, $this->repositoryId, $anchor);
 		$this->catalog->load(
-			fn () => array_map('basename', $this->backend->list('snapshots')),
-			fn () => array_map('basename', $this->backend->list('trash/info')),
+			fn () => array_map('basename', $this->backend->listFiles('snapshots')),
+			fn () => array_map('basename', $this->backend->listFiles('trash/info')),
 			fn () => $this->backend->list('users'));
 	}
 
@@ -262,6 +262,15 @@ final class Repository {
 		});
 	}
 
+	/** See BlobIndex::holdCache(): no database writes until releaseIndexCache(). */
+	public function holdIndexCache(): void {
+		$this->index->holdCache();
+	}
+
+	public function releaseIndexCache(): void {
+		$this->index->releaseCache();
+	}
+
 	public function flushPacks(): void {
 		$this->runPacks?->flush();
 	}
@@ -634,7 +643,7 @@ final class Repository {
 	 * @return list<string>
 	 */
 	public function listSnapshots(): array {
-		$present = array_flip(array_map('basename', $this->backend->list('snapshots')));
+		$present = array_flip(array_map('basename', $this->backend->listFiles('snapshots')));
 		$listed = array_keys($this->catalog->snapshots);
 		foreach ($listed as $id) {
 			if (!isset($present[$id])) {

tests/Unit/Backend/StorageBackendListTest.php

@@ -0,0 +1,57 @@
+<?php
+
+declare(strict_types=1);
+
+namespace OCA\NgBackup\Tests\Unit\Backend;
+
+use OCA\NgBackup\Backend\BackendException;
+use OCA\NgBackup\Backend\StorageBackend;
+use OCA\NgBackup\Tests\Unit\TempDirTrait;
+use OCP\Files\Storage\IStorage;
+use PHPUnit\Framework\TestCase;
+
+/** On a remote storage every is_dir() is a round trip: listing index/ must not cost one per pack. */
+class StorageBackendListTest extends TestCase {
+	use TempDirTrait;
+
+	public function testListFilesChecksOnlyTheFolderItself(): void {
+		$dir = $this->tempDir();
+		foreach (['b1', 'a2', 'c3.part-0123abcd'] as $name) {
+			touch("$dir/$name");
+		}
+		$isDirCalls = 0;
+		$storage = $this->createMock(IStorage::class);
+		$storage->method('instanceOfStorage')->willReturn(false);
+		$storage->method('is_dir')->willReturnCallback(function (string $path) use (&$isDirCalls): bool {
+			$isDirCalls++;
+			return $path === 'repo' || $path === 'repo/index';
+		});
+		$storage->method('opendir')->willReturnCallback(fn (string $path) => $path === 'repo/index' ? opendir($dir) : false);
+
+		$backend = new StorageBackend($storage, 'repo');
+		$isDirCalls = 0; // the constructor checks the base folder
+		$this->assertSame(['index/a2', 'index/b1'], $backend->listFiles('index'));
+		$this->assertSame(1, $isDirCalls);
+
+		$isDirCalls = 0;
+		$this->assertSame(['index/a2', 'index/b1'], $backend->list('index'));
+		$this->assertSame(4, $isDirCalls, 'list() checks every entry: the reason listFiles() exists');
+	}
+
+	public function testListFilesOfAMissingFolderIsEmpty(): void {
+		$storage = $this->createMock(IStorage::class);
+		$storage->method('instanceOfStorage')->willReturn(false);
+		$storage->method('is_dir')->willReturnCallback(fn (string $path): bool => $path === 'repo');
+		$storage->expects($this->never())->method('opendir');
+		$this->assertSame([], (new StorageBackend($storage, 'repo'))->listFiles('snapshots'));
+	}
+
+	public function testAnUnreadableFolderIsAnErrorNotEmpty(): void {
+		$storage = $this->createMock(IStorage::class);
+		$storage->method('instanceOfStorage')->willReturn(false);
+		$storage->method('is_dir')->willReturn(true);
+		$storage->method('opendir')->willReturn(false);
+		$this->expectException(BackendException::class);
+		(new StorageBackend($storage, 'repo'))->listFiles('catalog');
+	}
+}

tests/Unit/Repository/IndexCacheTest.php

@@ -51,4 +51,41 @@ class IndexCacheTest extends TestCase {
 		$fourth->restore($r1['snapshot'], $target);
 		$this->assertCount(12, self::hashTree($target));
 	}
+
+	/** The database dump reads in a read-only transaction: a pack that fills up then must not write the cache yet. */
+	public function testHeldCacheWritesWaitForRelease(): void {
+		$cache = new class implements IIndexCache {
+			public array $data = [];
+			public bool $readOnly = false;
+			public function all(string $r): array { return $this->data[$r] ?? []; }
+			public function put(string $r, string $p, array $e): void {
+				if ($this->readOnly) {
+					throw new \RuntimeException('cannot execute INSERT in a read-only transaction');
+				}
+				$this->data[$r][$p] = $e;
+			}
+			public function remove(string $r, array $ps): void { foreach ($ps as $p) { unset($this->data[$r][$p]); } }
+		};
+		$backend = new LocalBackend($this->tempDir());
+		$keys = KeyRing::generate();
+		Repository::initWithKey($backend, $keys, $keys->wrap('pw'));
+		$repo = Repository::openWithKey($backend, $keys, $cache);
+
+		$stats = [];
+		$repo->holdIndexCache();
+		$cache->readOnly = true;
+		$writer = $repo->blobWriter($stats);
+		for ($i = 0; $i < 40; $i++) {
+			$writer->write(random_bytes(1048576)); // more than one 32 MiB pack
+		}
+		$writer->finish();
+		$this->assertGreaterThan(0, count($backend->list('index')), 'a full pack was uploaded during the hold');
+		$this->assertSame([], $cache->data);
+
+		$cache->readOnly = false;
+		$repo->releaseIndexCache();
+		$repo->flushPacks();
+		$this->assertCount(count($backend->list('index')), array_merge(...array_values($cache->data)));
+		$this->assertSame(0, Repository::openWithKey($backend, $keys, $cache)->indexDownloads());
+	}
 }

tests/Unit/Repository/VerifyTest.php

@@ -166,6 +166,10 @@ final class FlakyRangeBackend implements IBackend {
 		return $this->inner->list($prefix);
 	}
 
+	public function listFiles(string $dir): array {
+		return $this->inner->listFiles($dir);
+	}
+
 	public function delete(string $path): void {
 		$this->inner->delete($path);
 	}

tests/bootstrap.php

@@ -3,3 +3,13 @@
 declare(strict_types=1);
 
 require __DIR__ . '/../vendor/autoload.php';
+
+// nextcloud/ocp ships the public interfaces without an autoloader; tests that mock them need one.
+spl_autoload_register(static function (string $class): void {
+	if (str_starts_with($class, 'OCP\\')) {
+		$file = __DIR__ . '/../vendor/nextcloud/ocp/' . str_replace('\\', '/', $class) . '.php';
+		if (is_file($file)) {
+			require $file;
+		}
+	}
+});