Compare commits

..

15 Commits

Author SHA1 Message Date
Dillard Blom 54750e9742 Merge pull request #8: Release 0.9.4-beta.5
CI / unit (8.2) (push) Has been cancelled
CI / unit (8.3) (push) Has been cancelled
CI / unit (8.4) (push) Has been cancelled
CI / psalm (push) Has been cancelled
CI / integration (mariadb, 32) (push) Has been cancelled
CI / integration (mariadb, 33) (push) Has been cancelled
CI / integration (mariadb, 34) (push) Has been cancelled
CI / integration (mariadb, 35) (push) Has been cancelled
CI / integration (pgsql, 32) (push) Has been cancelled
CI / integration (pgsql, 33) (push) Has been cancelled
CI / integration (pgsql, 34) (push) Has been cancelled
CI / integration (pgsql, 35) (push) Has been cancelled
CI / integration (sqlite, 32) (push) Has been cancelled
CI / integration (sqlite, 33) (push) Has been cancelled
CI / integration (sqlite, 34) (push) Has been cancelled
CI / integration (sqlite, 35) (push) Has been cancelled
CI / integration-targets (push) Has been cancelled
CI / integration-user (32) (push) Has been cancelled
CI / integration-user (33) (push) Has been cancelled
CI / integration-user (34) (push) Has been cancelled
CI / integration-user (35) (push) Has been cancelled
CI / integration-full-restore (push) Has been cancelled
Restore and deep verify: read a pack as one stream; pipelined SFTP reads
2026-10-09 12:05:31 +02:00
dillard fe92ef70bf Release 0.9.4-beta.5 2026-10-09 11:58:52 +02:00
dillard b3e3054737 SftpReadStream: fall back to the storage's own stream instead of failing a read 2026-10-09 11:24:37 +02:00
dillard c775e27dd2 CHANGELOG: SFTP reads with several requests in flight 2026-10-09 09:59:10 +02:00
dillard d8e772137f StorageBackend: read SFTP locations with several requests in flight
files_external's SFTP read stream asks for the next 256 KiB only after
the previous answer has arrived, so reading costs one round trip per
request whatever the line speed: 4.9 MB/s at 47 ms to a Hetzner Storage
Box. SftpReadStream keeps up to eight READ requests in flight over the
same phpseclib connection (15 MB/s on that link), starting with one so
a small read does not fetch megabytes ahead.

A server that answers less than asked is handled by asking again from
where the answer ended; when more data follows, that size is used for
later reads over the same connection, never below the 32 KiB every
server must support. Without the needed connection methods, or when the
open fails, the storage's own stream is used as before.
2026-10-09 04:27:21 +02:00
dillard bcfec17cfe README: WebDAV request timeout on slow uploads; deep verify no longer fetches per blob 2026-10-08 22:38:32 +02:00
dillard b309081d1a Repository: close the pack stream before any other request on the location
Review notes on the open pack stream: it stayed open while a backup run went on to upload
(e.g. the snapshot after reading the file list back), which SFTP cannot do on one connection.
Reading moves to PackReader, and Repository reaches the location through PackReaderGuard,
which closes the stream before every other request. A pack stream that cannot seek is read up
to the offset. Deep verify no longer reports an \Error during decoding as damage.
2026-10-08 22:34:02 +02:00
dillard 5f8d5faee7 Repository: read the blobs of a pack from one open stream
Restores and deep verifies requested every blob separately: a new file handle and a seek on
the location each time. They read the blobs of a snapshot in the order they were written, so
the stream of the pack read last now stays open and is read on; a jump backwards, to another
pack, or more than 8 MiB ahead opens the pack again at the offset, as before. On a Hetzner
Storage Box over SMB, restoring 86 MB in 202 files went from 0.78 MB/s to 4.2 MB/s.
2026-10-08 22:17:53 +02:00
Dillard Blom 4fa7cbf6ef Merge pull request #7: Release 0.9.3-beta.4
CI / unit (8.2) (push) Has been cancelled
CI / unit (8.3) (push) Has been cancelled
CI / unit (8.4) (push) Has been cancelled
CI / psalm (push) Has been cancelled
CI / integration (mariadb, 32) (push) Has been cancelled
CI / integration (mariadb, 33) (push) Has been cancelled
CI / integration (mariadb, 34) (push) Has been cancelled
CI / integration (mariadb, 35) (push) Has been cancelled
CI / integration (pgsql, 32) (push) Has been cancelled
CI / integration (pgsql, 33) (push) Has been cancelled
CI / integration (pgsql, 34) (push) Has been cancelled
CI / integration (pgsql, 35) (push) Has been cancelled
CI / integration (sqlite, 32) (push) Has been cancelled
CI / integration (sqlite, 33) (push) Has been cancelled
CI / integration (sqlite, 34) (push) Has been cancelled
CI / integration (sqlite, 35) (push) Has been cancelled
CI / integration-targets (push) Has been cancelled
CI / integration-user (32) (push) Has been cancelled
CI / integration-user (33) (push) Has been cancelled
CI / integration-user (34) (push) Has been cancelled
CI / integration-user (35) (push) Has been cancelled
CI / integration-full-restore (push) Has been cancelled
Release 0.9.3-beta.4
2026-10-08 21:18:00 +02:00
dillard f4abfe0a8d Release 0.9.3-beta.4 2026-10-08 20:46:14 +02:00
dillard fe1a289c2d Backend: an existing folder that cannot be listed is an error, not empty
Read as empty, a catalog/ or index/ folder would make an existing repository look new.
2026-10-08 20:46:14 +02:00
dillard ab3265e42c Backup: release the held index cache writes also when the dump transaction fails to start
holdIndexCache() ran before beginTransaction(), outside the try, and the release came after
rollBack(); an exception from either left later cache writes queued for the rest of the run.
2026-10-08 20:38:25 +02:00
dillard 466cff8857 CHANGELOG: fixes for the next hotfix 2026-10-08 18:28:29 +02:00
dillard 6bba0b59fc Repository: list flat folders without a type check per entry
Opening a repository lists index/, which holds one file per pack.
StorageBackend::list() checked every entry with is_dir(), and on a
remote storage each check is a round trip. Over SFTP to a Hetzner Storage
Box (33 ms) with a 60 GiB repository, occ backup:run (30 s steps) spent
nearly every step on these requests and uploaded 45 KiB/s; with this
change it uploads 5.8 MB/s. The cost grows with the number of packs.

index/, snapshots/, catalog/ and trash/info/ only ever hold objects:
list them with the new IBackend::listFiles(), one listing without
per-entry checks.
2026-10-08 18:28:29 +02:00
dillard 62766e1e71 Backup: do not write the index cache inside the read-only dump transaction
A database dump that fills a whole pack (32 MiB) uploads that pack while
the dump transaction is still open, and the pack's index cache row was
inserted on the same connection. PostgreSQL refuses that (25006, read-only
transaction) and the run failed; on other databases the row was rolled
back with the dump transaction. The index now holds those cache writes
back until the transaction is over.
2026-10-08 13:03:52 +02:00
19 changed files with 1546 additions and 54 deletions
+34
View File
@@ -7,6 +7,38 @@ 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.
## [0.9.4-beta.5] - 2026-10-09
Bugfix release: faster restores.
### Fixed
- Restores and `occ backup:verify --deep` fetched every blob with its own request, so on a
remote location they ran far below the speed of the connection. The blobs of a pack are now
read from one open stream, in the order they were written. Restoring 86 MB in 202 files from a
Hetzner Storage Box over SMB went from 0.78 MB/s to 4.2 MB/s, the speed of reading the packs
as plain files over that connection.
- Reading from an SFTP location waited a full round trip for every 256 KiB, whatever the line
speed. Reads now keep up to 8 requests in flight. Restoring 1.97 GB in 500 files from a Hetzner
Storage Box (47 ms away) took 694 s before the fix above, 499 s with it and 142 s with both
(2.8 → 3.9 → 13.9 MB/s); a full 51 GB restore ran at 13.0 MB/s. When a server error
interrupts such a read, it goes on from the same position with Nextcloud's own SFTP stream;
after three failing files in a row, or on a Nextcloud whose SFTP library lacks what this needs,
only Nextcloud's own stream is used.
- README: a WebDAV location on a slow upload needs a higher `davstorage.request_timeout`.
## [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 +154,8 @@ 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).
[0.9.4-beta.5]: https://github.com/dillardblom/ng_backup/releases/tag/v0.9.4-beta.5
[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
+5 -4
View File
@@ -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 (0.9.4).** 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.
@@ -136,9 +136,10 @@ Contributions welcome via PR — add what you ran into with your own provider.
Conflict`. SMB: share = the sub-account name, port 445 must be reachable from your server (it was
from a home connection here, but some providers block it), and the `smbclient` PECL extension is
required (see the SMB entry below). Both have a noticeable fixed cost per run: an incremental run
that uploads nothing took about 20 s over WebDAV and 50 s over SMB. A deep verify is slow over
remote storage, since each blob is fetched with its own request (19 minutes for 66 MiB over
WebDAV); run it occasionally, not after every backup.
that uploads nothing took about 20 s over WebDAV and 50 s over SMB. Over WebDAV, Nextcloud gives
every request at most 30 s by default, also an upload of a whole 32 MiB pack: on an upload
slower than about 9 Mbit/s a backup fails with `cURL error 28: Operation timed out`. Raise the
limit with `occ config:system:set davstorage.request_timeout --value=600 --type=integer`.
- **FTP, server-dependent absolute-path handling:** files_external's FTP backend always builds
absolute paths (e.g. `MKD /repo`) for the configured root, never relative ones. `atmoz/sftp`-style
+1 -1
View File
@@ -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>0.9.4</version>
<licence>agpl</licence>
<author>Dillard Blom</author>
<namespace>NgBackup</namespace>
+10
View File
@@ -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;
}
+19
View File
@@ -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));
}
+412
View File
@@ -0,0 +1,412 @@
<?php
declare(strict_types=1);
// SPDX-FileCopyrightText: 2026 Dillard Blom
// SPDX-License-Identifier: AGPL-3.0-or-later
namespace OCA\NgBackup\Backend;
/**
* Read stream for files_external's SFTP storage that keeps several READ requests in flight.
*
* files_external's own read stream asks for the next 256 KiB only after the previous one has
* arrived, so every request costs a full round trip: about 5 MB/s at 47 ms to a Hetzner Storage
* Box, whatever the line speed. With 8 requests in flight the same connection reached 15 MB/s.
* phpseclib's get() pipelines too, but only into memory or a local file, and only from offset 0.
*
* Talks to the storage's phpseclib connection through the same internal methods files_external's
* read stream uses (_send_sftp_packet, _get_sftp_packet, _realpath); open() returns null when the
* connection does not offer them (phpseclib 3 does not), and StorageBackend then falls back to the
* storage's own stream. A restore must rather be slow than fail: when a read fails halfway, the
* stream goes on from the same position with the storage's own stream. After MAX_FAILURES streams
* in a row failed, open() returns null for the rest of the process; one that works resets the count.
* Like that stream, it owns the connection while open: no other request may go over the same
* connection until it is closed (PackReaderGuard takes care of that for the repository).
*/
final class SftpReadStream {
private const PROTOCOL = 'ngbsftpread';
/** Requests in flight once a stream reads sequentially; it starts at one and doubles. */
public const DEPTH = 8;
/** OpenSSH and the Hetzner Storage Box both answer at most 255 KiB per READ. */
private const CHUNK = 261120;
/** Every SFTP server must handle packets of this size (draft-ietf-secsh-filexfer-02, section 3). */
private const MIN_CHUNK = 32768;
private const FXP_OPEN = 3;
private const FXP_CLOSE = 4;
private const FXP_READ = 5;
private const FXP_STATUS = 101;
private const FXP_HANDLE = 102;
private const FXP_DATA = 103;
private const FXF_READ = 1;
private const FX_EOF = 1;
/**
* READ size per connection, lowered when a server answers less than asked and then goes on, so
* later streams over that connection ask for no more than it gives.
*
* @var \WeakMap<object, int>|null
*/
private static ?\WeakMap $chunks = null;
private static int $requestId = 0x4e470000;
/** Streams in a row that failed; at MAX_FAILURES only the storage's own stream is used. */
private static int $failures = 0;
public const MAX_FAILURES = 3;
/** Set by stream_open() when the connection failed rather than the file. */
private static bool $openBroke = false;
/** @var resource|null set by PHP for stream wrappers */
public $context;
/**
* The phpseclib connection. PHP creates stream wrappers without constructor arguments; it is
* set in stream_open(), and no other method is called when that fails.
*
* @psalm-suppress PropertyNotSetInConstructor
*/
private object $sftp;
/** The storage and the path on it, to fall back to the storage's own stream. */
private ?object $storage = null;
private string $file = '';
/** @var resource|null the storage's own stream, once this one has failed */
private $fallback = null;
private string $handle = '';
/** Position of the reader: the first byte of $buffer. */
private int $pos = 0;
/** Offset of the next READ to send. */
private int $next = 0;
private string $buffer = '';
/** @var list<array{int, int, int}> request id, offset, length, in the order sent */
private array $inFlight = [];
private int $depth = 1;
private bool $eof = false;
/** This stream gave up on its own requests. */
private bool $failed = false;
/** The connection is in an unknown state: nothing more is sent over it or read from it. */
private bool $broken = false;
/** Length of the last short answer, until the next answer tells whether it was the end of the file. */
private ?int $shortAnswer = null;
/**
* @param object $storage files_external's SFTP storage (getConnection(), getRoot())
* @return resource|false|null the stream, false when the file cannot be opened, null when the
* connection does not offer what this stream needs
* @throws BackendException when the connection failed after the OPEN request went out: the
* storage's own stream must not be tried on it, as it would take a
* late answer to that request for its own
*/
public static function open(object $storage, string $path) {
if (self::$failures >= self::MAX_FAILURES) {
return null;
}
try {
if (!is_callable([$storage, 'getConnection']) || !is_callable([$storage, 'getRoot'])) {
return null;
}
$sftp = $storage->getConnection();
// is_callable and get_object_vars from here only see what is public: phpseclib 3 has
// methods of these names neither public nor at all.
if (!is_object($sftp) || !is_callable([$sftp, '_send_sftp_packet']) || !is_callable([$sftp, '_get_sftp_packet'])
|| !is_callable([$sftp, '_realpath']) || !array_key_exists('packet_type', get_object_vars($sftp))) {
return null;
}
if (!in_array(self::PROTOCOL, stream_get_wrappers(), true)) {
stream_wrapper_register(self::PROTOCOL, self::class);
}
$root = rtrim((string)$storage->getRoot(), '/');
$context = stream_context_create([self::PROTOCOL => [
'session' => $sftp,
'path' => $root . '/' . ltrim($path, '/'),
'storage' => $storage,
'file' => $path,
]]);
self::$openBroke = false;
$fh = @fopen(self::PROTOCOL . '://file', 'r', false, $context);
} catch (\Throwable) {
// nothing was sent yet
self::$failures++;
return null;
}
if ($fh === false && self::openBroke()) {
throw new BackendException('SFTP connection failed while opening ' . $path);
}
return $fh;
}
/** Test hook: forget the learned READ sizes and earlier failures. */
public static function reset(): void {
self::$chunks = null;
self::$failures = 0;
}
public function stream_open(string $url, string $mode, int $options, ?string &$openedPath): bool {
$options = stream_context_get_options($this->context ?? stream_context_get_default())[self::PROTOCOL] ?? [];
if (!isset($options['session'], $options['path']) || !is_object($options['session'])) {
return false;
}
$this->sftp = $options['session'];
$storage = $options['storage'] ?? null;
$this->storage = is_object($storage) ? $storage : null;
$this->file = (string)($options['file'] ?? '');
try {
$remote = $this->sftp->_realpath((string)$options['path']);
if (!is_string($remote)) {
return false;
}
$id = self::nextId();
if (!$this->sftp->_send_sftp_packet(self::FXP_OPEN, pack('Na*N2', strlen($remote), $remote, self::FXF_READ, 0), $id)) {
return $this->connectionFailed();
}
$response = $this->sftp->_get_sftp_packet($id);
} catch (\Throwable) {
return $this->connectionFailed();
}
if (!is_string($response)) {
return $this->connectionFailed();
}
if ($this->sftp->packet_type !== self::FXP_HANDLE) {
return false; // an answer, but no handle: the file cannot be opened
}
$this->handle = substr($response, 4);
return true;
}
/** Not a missing file but a connection that does not behave as expected. */
private function connectionFailed(): bool {
self::$failures++;
self::$openBroke = true;
return false;
}
/** Whether the last stream_open() failed on the connection; a call, as fopen() sets it. */
private static function openBroke(): bool {
return self::$openBroke;
}
public function stream_read(int $count): string|false {
if ($this->fallback !== null) {
return fread($this->fallback, $count);
}
while (strlen($this->buffer) < $count && !$this->eof) {
if (!$this->receive()) {
return $this->useFallback() ? $this->stream_read($count) : false;
}
}
$data = substr($this->buffer, 0, $count);
$this->buffer = substr($this->buffer, strlen($data));
$this->pos += strlen($data);
return $data;
}
public function stream_eof(): bool {
if ($this->fallback !== null) {
return feof($this->fallback);
}
return $this->eof && $this->buffer === '';
}
public function stream_tell(): int {
if ($this->fallback !== null) {
$pos = ftell($this->fallback);
return $pos === false ? $this->pos : $pos;
}
return $this->pos;
}
public function stream_seek(int $offset, int $whence = SEEK_SET): bool {
if ($this->fallback !== null) {
return fseek($this->fallback, $offset, $whence) === 0;
}
if ($whence === SEEK_CUR) {
$offset += $this->pos;
} elseif ($whence !== SEEK_SET) {
return false; // the size is never asked for, to save a round trip
}
if ($offset < 0) {
return false;
}
if ($this->failed) {
$this->pos = $offset;
return $this->useFallback();
}
if ($offset >= $this->pos && $offset - $this->pos <= strlen($this->buffer)) {
$this->buffer = substr($this->buffer, $offset - $this->pos);
} else {
$this->drain();
$this->buffer = '';
$this->next = $offset;
$this->eof = false;
$this->depth = 1;
$this->shortAnswer = null;
}
$this->pos = $offset;
return true;
}
/** @return array<int|string, int> */
public function stream_stat(): array {
return [];
}
public function stream_close(): void {
if (!$this->failed) {
self::$failures = 0;
}
$fallback = $this->fallback;
$this->fallback = null;
if ($fallback !== null) {
fclose($fallback);
}
$this->closeHandle();
}
/**
* Go on with the storage's own stream from the current position, after a server error on this
* one. The data still buffered is dropped and read again. False after a connection error, or
* when that stream cannot be opened or positioned: then the read fails as it would have
* without this class. Reopening by path is safe for a repository: packs and index files are
* never rewritten, and every object is authenticated when it is decrypted.
*/
private function useFallback(): bool {
$this->closeHandle();
$this->buffer = '';
// files_external's stream takes whatever answer comes next on the connection: after a
// connection error, answers to this stream's requests may still be on their way.
if ($this->broken || $this->storage === null || !is_callable([$this->storage, 'fopen'])) {
return false;
}
try {
$fh = $this->storage->fopen($this->file, 'r');
if (!is_resource($fh)) {
return false;
}
if ($this->pos > 0 && fseek($fh, $this->pos) !== 0) {
fclose($fh);
return false;
}
} catch (\Throwable) {
return false;
}
$this->fallback = $fh;
return true;
}
private function closeHandle(): void {
$this->drain();
if ($this->handle !== '' && !$this->broken) {
try {
$id = self::nextId();
if (!$this->sftp->_send_sftp_packet(self::FXP_CLOSE, pack('Na*', strlen($this->handle), $this->handle), $id)
|| !is_string($this->sftp->_get_sftp_packet($id))) {
$this->broken = true;
}
} catch (\Throwable) {
$this->broken = true;
}
}
$this->handle = '';
}
/** Keep the pipeline full, then take the oldest answer. False on a connection or server error. */
private function receive(): bool {
if ($this->failed) {
return false;
}
try {
return $this->receiveOne();
} catch (\Throwable) {
$this->broken = true;
return $this->fail();
}
}
private function receiveOne(): bool {
while (count($this->inFlight) < $this->depth) {
$id = self::nextId();
$length = self::chunk($this->sftp);
$packet = pack('Na*N3', strlen($this->handle), $this->handle, intdiv($this->next, 4294967296), $this->next % 4294967296, $length);
if (!$this->sftp->_send_sftp_packet(self::FXP_READ, $packet, $id)) {
$this->broken = true;
return $this->fail();
}
$this->inFlight[] = [$id, $this->next, $length];
$this->next += $length;
}
[$id, $offset, $length] = array_shift($this->inFlight);
$response = $this->sftp->_get_sftp_packet($id);
if (!is_string($response)) {
$this->broken = true;
return $this->fail();
}
if ($this->sftp->packet_type === self::FXP_DATA) {
$data = substr($response, 4);
if ($data === '') {
return $this->fail();
}
$this->buffer .= $data;
$this->depth = min(self::DEPTH, $this->depth * 2);
if ($this->shortAnswer !== null) {
// More data after a short answer: that one was the server's cap, not the end.
// Never below what every server must support: a one-off small answer is no cap.
self::$chunks ??= new \WeakMap();
self::$chunks[$this->sftp] = max(self::MIN_CHUNK, min(self::chunk($this->sftp), $this->shortAnswer));
$this->shortAnswer = null;
}
if (strlen($data) < $length) {
// Fewer bytes than asked: the end of the file, or a server that caps a READ. The
// requests behind this one assumed a full answer, so drop them and ask again from here.
$this->drain();
$this->next = $offset + strlen($data);
$this->shortAnswer = strlen($data);
}
return true;
}
$status = $this->sftp->packet_type === self::FXP_STATUS ? unpack('N', $response) : false;
if (is_array($status) && ($status[1] ?? null) === self::FX_EOF) {
$this->eof = true;
$this->shortAnswer = null;
$this->drain();
return true;
}
return $this->fail();
}
private function fail(): bool {
$this->drain();
if (!$this->failed) {
self::$failures++;
}
$this->failed = true;
return false;
}
/** Read and drop the answers still on their way, so the connection is free again. */
private function drain(): void {
$inFlight = $this->inFlight;
$this->inFlight = [];
if ($this->broken) {
return;
}
try {
foreach ($inFlight as [$id]) {
if (!is_string($this->sftp->_get_sftp_packet($id))) {
$this->broken = true;
return;
}
}
} catch (\Throwable) {
$this->broken = true;
}
}
private static function chunk(object $sftp): int {
$learned = self::$chunks?->offsetExists($sftp) === true ? self::$chunks->offsetGet($sftp) : null;
return is_int($learned) ? $learned : self::CHUNK;
}
private static function nextId(): int {
self::$requestId = self::$requestId >= 0x7fffffff ? 0x4e470000 : self::$requestId + 1;
return self::$requestId;
}
}
+33
View File
@@ -17,6 +17,8 @@ use OCP\Files\Storage\IWriteStreamStorage;
final class StorageBackend implements IBackend {
/** Object stores write whole objects atomically; other storages get a temp name + rename. */
private bool $atomicPut;
/** SFTP: read through SftpReadStream, which keeps several requests in flight. */
private bool $sftpReads;
public function __construct(
private IStorage $storage,
@@ -25,6 +27,7 @@ final class StorageBackend implements IBackend {
$this->base = trim($base, '/');
self::validate($this->base, true);
$this->atomicPut = $storage->instanceOfStorage(\OCA\Files_External\Lib\Storage\AmazonS3::class);
$this->sftpReads = $storage->instanceOfStorage(\OCA\Files_External\Lib\Storage\SFTP::class);
if ($this->base !== '') {
$this->mkdirs($this->base);
}
@@ -79,6 +82,14 @@ final class StorageBackend implements IBackend {
}
public function get(string $path) {
if ($this->sftpReads) {
$fh = SftpReadStream::open($this->storage, $this->abs($path));
if (is_resource($fh)) {
return $fh;
}
// null: not supported or failing for now; false: not opened, which need not mean
// missing. Either way the storage's own stream decides.
}
$fh = $this->storage->fopen($this->abs($path), 'r');
if ($fh === false) {
throw new BackendException('Not found: ' . $path);
@@ -133,6 +144,28 @@ 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;
+43 -36
View File
@@ -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;
+45 -3
View File
@@ -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[$id]);
+1 -1
View File
@@ -128,7 +128,7 @@ final class Catalog {
/** @return list<int> */
private function generations(): array {
return array_map(fn ($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 {
+97
View File
@@ -0,0 +1,97 @@
<?php
declare(strict_types=1);
// SPDX-FileCopyrightText: 2026 Dillard Blom
// SPDX-License-Identifier: AGPL-3.0-or-later
namespace OCA\NgBackup\Repository;
use OCA\NgBackup\Backend\IBackend;
/**
* Reads blobs from packs. Restores and deep verifies read the blobs of a snapshot in the order
* they were written, so the stream of the pack read last is kept open and read on from where it
* is: one request per pack instead of one per blob (a round trip each, and on most remote
* storages a new connection or file handle too). Anything else opens the pack at the offset,
* as a single read always did.
*
* The open stream must not overlap other requests on the location (SFTP cannot read and write
* on one connection at once): Repository reaches the location through a PackReaderGuard, which
* closes it first.
*/
final class PackReader {
/** A gap up to this size in the open stream is read past instead of opening the pack again. */
private const SKIP_MAX = 8 * 1048576;
private const CHUNK = 1048576;
/** @var resource|null */
private $stream = null;
private string $pack = '';
private int $pos = 0;
public function __construct(
private IBackend $backend,
) {
}
public function read(string $pack, int $offset, int $length): string {
if ($this->stream !== null && $this->pack === $pack && $offset >= $this->pos && $offset - $this->pos <= self::SKIP_MAX) {
try {
$this->skipTo($offset);
return $this->readExactly($length);
} catch (\Exception) {
// The open stream broke (e.g. a timeout); open the pack again below.
}
}
$this->close();
$fh = $this->backend->get('packs/' . substr($pack, 0, 2) . '/' . $pack);
$this->stream = $fh;
$this->pack = $pack;
$this->pos = 0;
try {
if ($offset > 0 && (stream_get_meta_data($fh)['seekable'] ?? false) && fseek($fh, $offset) === 0) {
$this->pos = $offset;
}
$this->skipTo($offset); // a stream that cannot seek is read up to the offset
return $this->readExactly($length);
} catch (\Exception $e) {
$this->close();
throw $e;
}
}
public function close(): void {
$fh = $this->stream;
$this->stream = null;
if ($fh !== null) {
@fclose($fh);
}
}
public function __destruct() {
$this->close();
}
private function skipTo(int $offset): void {
while ($this->pos < $offset) {
$this->readExactly(min($offset - $this->pos, self::CHUNK));
}
}
/** Read $length bytes from the open stream; a short read closes it. */
private function readExactly(int $length): string {
$fh = $this->stream ?? throw new RepositoryException('No open pack stream');
$data = '';
while (strlen($data) < $length) {
$chunk = fread($fh, min($length - strlen($data), self::CHUNK));
if ($chunk === false || $chunk === '') {
$this->close();
throw new RepositoryException('Short read in pack ' . $this->pack);
}
$data .= $chunk;
}
$this->pos += $length;
return $data;
}
}
+60
View File
@@ -0,0 +1,60 @@
<?php
declare(strict_types=1);
// SPDX-FileCopyrightText: 2026 Dillard Blom
// SPDX-License-Identifier: AGPL-3.0-or-later
namespace OCA\NgBackup\Repository;
use OCA\NgBackup\Backend\IBackend;
/** The location as Repository sees it: closes the PackReader's open stream before every other request. */
final class PackReaderGuard implements IBackend {
public function __construct(
private IBackend $inner,
private PackReader $reader,
) {
}
/** @param resource $stream */
public function put(string $path, $stream): void {
$this->reader->close();
$this->inner->put($path, $stream);
}
public function get(string $path) {
$this->reader->close();
return $this->inner->get($path);
}
public function getRange(string $path, int $offset, int $length): string {
$this->reader->close();
return $this->inner->getRange($path, $offset, $length);
}
public function exists(string $path): bool {
$this->reader->close();
return $this->inner->exists($path);
}
public function move(string $from, string $to): void {
$this->reader->close();
$this->inner->move($from, $to);
}
public function list(string $prefix): array {
$this->reader->close();
return $this->inner->list($prefix);
}
public function listFiles(string $dir): array {
$this->reader->close();
return $this->inner->listFiles($dir);
}
public function delete(string $path): void {
$this->reader->close();
$this->inner->delete($path);
}
}
+19 -7
View File
@@ -36,6 +36,7 @@ final class Repository {
private ?PackWriter $runPacks = null;
private ?int $maxBytes = null;
private Catalog $catalog;
private PackReader $packReader;
private function __construct(
private IBackend $backend,
@@ -43,7 +44,9 @@ final class Repository {
private string $repositoryId,
) {
$this->cipher = new StreamCipher($keys);
$this->index = new BlobIndex($backend, $this->cipher);
$this->packReader = new PackReader($backend);
$this->backend = new PackReaderGuard($backend, $this->packReader);
$this->index = new BlobIndex($this->backend, $this->cipher);
}
public static function init(IBackend $backend, #[\SensitiveParameter] string $passphrase): self {
@@ -92,8 +95,8 @@ 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 +265,15 @@ 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 +646,7 @@ 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])) {
@@ -699,7 +711,7 @@ final class Repository {
private function loadBlob(string $id): string {
[$pack, $offset, $length] = $this->index->get($id);
return $this->decodeBlob($id, $this->backend->getRange('packs/' . substr($pack, 0, 2) . '/' . $pack, $offset, $length));
return $this->decodeBlob($id, $this->packReader->read($pack, $offset, $length));
}
/** Decrypts and checks a blob read from its pack; throws when it is corrupted or tampered with. */
@@ -839,7 +851,7 @@ final class Repository {
$encrypted = null;
for ($attempt = 1; $attempt <= self::VERIFY_READ_ATTEMPTS && $encrypted === null; $attempt++) {
try {
$encrypted = $this->backend->getRange('packs/' . substr($pack, 0, 2) . '/' . $pack, $offset, $length);
$encrypted = $this->packReader->read($pack, $offset, $length);
} catch (\Exception $e) { // storage adapters throw their own exceptions; \Error (a bug) is not retried
$stats['readError'] = $e->getMessage();
}
@@ -850,7 +862,7 @@ final class Repository {
}
try {
$stats['bytesChecked'] += strlen($this->decodeBlob($id, $encrypted));
} catch (\Throwable) {
} catch (\Exception) { // damage; an \Error is a bug and must not be reported as damage
$stats['failed'][] = $id;
}
}
+478
View File
@@ -0,0 +1,478 @@
<?php
declare(strict_types=1);
namespace OCA\NgBackup\Tests\Unit\Backend;
use OCA\NgBackup\Backend\BackendException;
use OCA\NgBackup\Backend\SftpReadStream;
use PHPUnit\Framework\TestCase;
/**
* SftpReadStream against a fake of the phpseclib connection: an SFTP server that answers READ
* requests from in-memory files, optionally capped. Every answer must be asked for by its request
* id, as phpseclib matches them: a server may answer out of order.
*/
class SftpReadStreamTest extends TestCase {
protected function setUp(): void {
SftpReadStream::reset();
}
public function testReadsAWholeFileWithSeveralRequestsInFlight(): void {
$data = random_bytes(5 * 1048576 + 12345);
$sftp = new FakeSftpConnection(['/home/repo/packs/ab/abcd' => $data]);
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'repo/packs/ab/abcd');
$this->assertIsResource($fh);
$this->assertSame($data, stream_get_contents($fh));
$this->assertTrue(feof($fh));
fclose($fh);
$this->assertSame(SftpReadStream::DEPTH, $sftp->maxInFlight);
$this->assertSame(0, $sftp->pending(), 'nothing may be left on the connection after close');
$this->assertSame([], $sftp->openHandles);
}
public function testAServerThatCapsAReadStillGivesEveryByteInOrder(): void {
$data = random_bytes(3 * 1048576 + 7);
$sftp = new FakeSftpConnection(['/home/f' => $data], cap: 100000);
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'f');
$this->assertSame($data, stream_get_contents($fh));
fclose($fh);
$this->assertSame(0, $sftp->pending());
// The cap is learned: a next stream asks for no more than the server gives.
$sftp->reads = [];
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'f');
$this->assertSame($data, stream_get_contents($fh));
fclose($fh);
$this->assertSame([100000], array_values(array_unique($sftp->reads)));
// Only for that connection.
$other = new FakeSftpConnection(['/home/f' => $data]);
$fh = SftpReadStream::open(new FakeSftpStorage($other), 'f');
$this->assertSame($data, stream_get_contents($fh));
fclose($fh);
$this->assertSame(261120, max($other->reads));
}
public function testATinyAnswerDoesNotShrinkTheReadsBelowWhatEveryServerSupports(): void {
$data = random_bytes(1048576);
$sftp = new FakeSftpConnection(['/home/f' => $data], cap: 1000);
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'f');
$this->assertSame($data, stream_get_contents($fh));
fclose($fh);
$this->assertSame(32768, min($sftp->reads));
}
public function testSeekingReadsFromTheNewOffset(): void {
$data = random_bytes(3 * 1048576);
$sftp = new FakeSftpConnection(['/home/f' => $data]);
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'f');
$this->assertSame(0, fseek($fh, 2000000));
$this->assertSame(2000000, ftell($fh));
$this->assertSame(substr($data, 2000000, 1000), fread($fh, 1000));
$this->assertSame(0, fseek($fh, 10)); // backwards
$this->assertSame(substr($data, 10, 70000), $this->readExactly($fh, 70000));
$this->assertSame(0, fseek($fh, 100, SEEK_CUR)); // within what is already buffered
$this->assertSame(substr($data, 70110, 5000), $this->readExactly($fh, 5000));
fclose($fh);
$this->assertSame(0, $sftp->pending());
}
public function testASmallReadDoesNotFetchTheWholePipelineAhead(): void {
$sftp = new FakeSftpConnection(['/home/f' => random_bytes(10 * 1048576)]);
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'f');
fread($fh, 100);
fclose($fh);
$this->assertCount(1, $sftp->reads, 'the first request goes alone, the pipeline grows while reading on');
}
public function testAnEmptyFileIsAtEndAtOnce(): void {
$sftp = new FakeSftpConnection(['/home/f' => '']);
$fh = SftpReadStream::open(new FakeSftpStorage($sftp), 'f');
$this->assertSame('', stream_get_contents($fh));
$this->assertTrue(feof($fh));
fclose($fh);
}
public function testAMissingFileCannotBeOpened(): void {
$sftp = new FakeSftpConnection([]);
$this->assertFalse(SftpReadStream::open(new FakeSftpStorage($sftp), 'nope'));
$this->assertSame(0, $sftp->pending());
}
public function testAServerErrorGoesOnWithTheStoragesOwnStreamFromTheSamePosition(): void {
$data = random_bytes(3 * 1048576);
$sftp = new FakeSftpConnection(['/home/f' => $data], failAt: 1048576);
$storage = new FakeSftpStorage($sftp);
$fh = SftpReadStream::open($storage, 'f');
$this->assertSame($data, $this->readAll($fh));
$this->assertSame(['f'], $storage->fopens);
$this->assertSame(strlen($data), ftell($fh));
fclose($fh);
$this->assertSame(0, $sftp->pending(), 'the connection is clean for the storage\'s own stream');
$this->assertSame([], $sftp->openHandles, 'the failed handle is closed');
// One failure is no reason to give up on the fast reads.
$fh = SftpReadStream::open($storage, 'f');
$this->assertIsResource($fh);
fclose($fh);
}
public function testOnlyFailuresInARowSwitchTheFastReadsOff(): void {
$data = random_bytes(2 * 1048576);
$failing = new FakeSftpStorage(new FakeSftpConnection(['/home/f' => $data], failAt: 1048576));
$working = new FakeSftpStorage(new FakeSftpConnection(['/home/f' => $data]));
for ($i = 1; $i < SftpReadStream::MAX_FAILURES; $i++) {
$fh = SftpReadStream::open($failing, 'f');
$this->assertSame($data, $this->readAll($fh));
fclose($fh);
}
$fh = SftpReadStream::open($working, 'f'); // a stream that works resets the count
$this->assertSame($data, $this->readAll($fh));
fclose($fh);
for ($i = 1; $i <= SftpReadStream::MAX_FAILURES; $i++) {
$fh = SftpReadStream::open($failing, 'f');
$this->assertIsResource($fh, "failure $i");
$this->assertSame($data, $this->readAll($fh));
fclose($fh);
}
$this->assertNull(SftpReadStream::open($working, 'f'));
$this->assertSame(SftpReadStream::MAX_FAILURES * 2 - 1, count($failing->fopens), 'every failed stream still gave all its data');
}
public function testAServerErrorAfterASeekGoesOnFromTheSeekedPosition(): void {
$data = random_bytes(3 * 1048576);
$sftp = new FakeSftpConnection(['/home/f' => $data], failAt: 2500000);
$storage = new FakeSftpStorage($sftp);
$fh = SftpReadStream::open($storage, 'f');
$this->assertSame(0, fseek($fh, 2400000));
$this->assertSame(substr($data, 2400000), $this->readAll($fh));
$this->assertSame(0, fseek($fh, 100)); // seeks go to the storage's stream now
$this->assertSame(substr($data, 100, 1000), $this->readExactly($fh, 1000));
fclose($fh);
}
public function testAConnectionErrorDoesNotFallBackOnTheSameConnection(): void {
// After an exception, answers may still be on their way; files_external's stream would
// take them for its own.
$sftp = new FakeSftpConnection(['/home/f' => random_bytes(3 * 1048576)], throwAt: 1048576);
$storage = new FakeSftpStorage($sftp);
$fh = SftpReadStream::open($storage, 'f');
$this->assertLessThan(3 * 1048576, strlen($this->readAll($fh)));
fclose($fh);
$this->assertSame([], $storage->fopens);
}
public function testALostConnectionDoesNotFallBackEither(): void {
$sftp = new FakeSftpConnection(['/home/f' => random_bytes(3 * 1048576)], loseAt: 1048576);
$storage = new FakeSftpStorage($sftp);
$fh = SftpReadStream::open($storage, 'f');
$this->assertLessThan(3 * 1048576, strlen($this->readAll($fh)));
fclose($fh);
$this->assertSame([], $storage->fopens);
}
public function testWhenTheStoragesOwnStreamFailsTooTheReadFails(): void {
$sftp = new FakeSftpConnection(['/home/f' => random_bytes(3 * 1048576)], failAt: 1048576);
$storage = new FakeSftpStorage($sftp, fopenFails: true);
$fh = SftpReadStream::open($storage, 'f');
$this->assertLessThan(3 * 1048576, strlen($this->readAll($fh)));
fclose($fh);
$this->assertSame(['f'], $storage->fopens);
}
public function testAConnectionErrorWhileOpeningIsAnErrorNotAMissingFile(): void {
// The OPEN request went out: the storage's own stream must not be tried on this connection.
$storage = new FakeSftpStorage(new FakeSftpConnection(['/home/f' => 'x'], throwAt: -1));
try {
SftpReadStream::open($storage, 'f');
$this->fail('expected a BackendException');
} catch (BackendException $e) {
$this->assertStringContainsString('connection', $e->getMessage());
}
$lost = new FakeSftpStorage(new FakeSftpConnection(['/home/f' => 'x'], loseAt: -1));
$this->expectException(BackendException::class);
SftpReadStream::open($lost, 'f');
}
public function testAConnectionErrorWhileOpeningDoesNotStopTheNextFile(): void {
try {
SftpReadStream::open(new FakeSftpStorage(new FakeSftpConnection(['/home/f' => 'x'], throwAt: -1)), 'f');
} catch (BackendException) {
}
$fh = SftpReadStream::open(new FakeSftpStorage(new FakeSftpConnection(['/home/f' => 'x'])), 'f');
$this->assertSame('x', stream_get_contents($fh));
fclose($fh);
}
public function testAFailedCloseDoesNotFallBackOnTheSameConnection(): void {
$sftp = new FakeSftpConnection(['/home/f' => random_bytes(3 * 1048576)], failAt: 1048576, closeFails: true);
$storage = new FakeSftpStorage($sftp);
$fh = SftpReadStream::open($storage, 'f');
$this->assertLessThan(3 * 1048576, strlen($this->readAll($fh)));
fclose($fh);
$this->assertSame([], $storage->fopens);
}
public function testAStorageThatThrowsIsNotUsed(): void {
$storage = new class {
public function getRoot(): string {
return '/';
}
public function getConnection(): object {
throw new \RuntimeException('Login failed');
}
};
$this->assertNull(SftpReadStream::open($storage, 'f'));
}
public function testAConnectionWithoutTheNeededMethodsIsNotUsed(): void {
$storage = new class {
public function getRoot(): string {
return '/';
}
public function getConnection(): object {
return new \stdClass();
}
};
$this->assertNull(SftpReadStream::open($storage, 'f'));
$this->assertNull(SftpReadStream::open(new \stdClass(), 'f'));
}
public function testAPhpseclib3ConnectionIsNotUsed(): void {
// phpseclib 3 has no _-prefixed methods; its packet_type and packet methods are private.
$v3 = new class {
private int $packet_type = -1;
private function _send_sftp_packet(): bool {
return true;
}
private function _get_sftp_packet(): string {
return '';
}
public function _realpath(string $path): string {
return $path;
}
public function realpath(string $path): string {
return $path . (string)$this->packet_type . $this->_send_sftp_packet() . $this->_get_sftp_packet();
}
};
$storage = new class($v3) {
public function __construct(
private object $connection,
) {
}
public function getRoot(): string {
return '/';
}
public function getConnection(): object {
return $this->connection;
}
};
$this->assertNull(SftpReadStream::open($storage, 'f'));
}
public function testAPrivatePacketTypeIsNotUsed(): void {
$conn = new class {
private int $packet_type = -1;
public function _send_sftp_packet(): bool {
return true;
}
public function _get_sftp_packet(): string {
return (string)$this->packet_type;
}
public function _realpath(string $path): string {
return $path;
}
};
$storage = new class($conn) {
public function __construct(
private object $connection,
) {
}
public function getRoot(): string {
return '/';
}
public function getConnection(): object {
return $this->connection;
}
};
$this->assertNull(SftpReadStream::open($storage, 'f'));
}
/** @param resource $fh */
private function readAll($fh): string {
$data = '';
while (!feof($fh)) {
$chunk = fread($fh, 65536);
if ($chunk === false || $chunk === '') {
break;
}
$data .= $chunk;
}
return $data;
}
/** @param resource $fh */
private function readExactly($fh, int $length): string {
$data = '';
while (strlen($data) < $length && !feof($fh)) {
$data .= fread($fh, $length - strlen($data));
}
return $data;
}
}
final class FakeSftpStorage {
/** @var list<string> paths opened through the storage's own stream */
public array $fopens = [];
public function __construct(
private FakeSftpConnection $connection,
private bool $fopenFails = false,
) {
}
/** files_external's own read stream: here the file from memory, seekable. */
public function fopen(string $path, string $mode) {
$this->fopens[] = $path;
$data = $this->connection->file('/home/' . $path);
if ($this->fopenFails || $data === null) {
return false;
}
$fh = fopen('php://memory', 'r+');
fwrite($fh, $data);
rewind($fh);
return $fh;
}
public function getRoot(): string {
return '/home/';
}
public function getConnection(): FakeSftpConnection {
return $this->connection;
}
}
/** The part of phpseclib\Net\SFTP (2.x) that SftpReadStream uses. */
final class FakeSftpConnection {
public int $packet_type = -1;
public int $maxInFlight = 0;
/** @var list<int> lengths asked for by READ requests */
public array $reads = [];
/** @var array<string, string> */
public array $openHandles = [];
/** @var array<int, array{int, string}> answers by request id, in the order the server sent them */
private array $answers = [];
private int $handles = 0;
/** @param array<string, string> $files */
public function __construct(
private array $files,
private int $cap = 261120,
private ?int $failAt = null,
/** READ at or past this offset throws; -1: OPEN throws */
private ?int $throwAt = null,
/** READ at or past this offset finds the connection gone (phpseclib returns false); -1: OPEN */
private ?int $loseAt = null,
private bool $closeFails = false,
) {
}
public function file(string $path): ?string {
return $this->files[$path] ?? null;
}
public function pending(): int {
return count($this->answers);
}
public function _realpath(string $path): string {
return '/' . trim(preg_replace('#/+#', '/', $path), '/');
}
public function _send_sftp_packet(int $type, string $data, int $id = 1): bool {
switch ($type) {
case 3: // OPEN
if ($this->throwAt === -1) {
throw new \RuntimeException('Connection closed by server');
}
if ($this->loseAt === -1) {
return false;
}
$path = substr($data, 4, unpack('N', $data)[1]);
if (!isset($this->files[$path])) {
$this->answers[$id] = [101, pack('N', 2)];
break;
}
$handle = 'h' . ++$this->handles;
$this->openHandles[$handle] = $path;
$this->answers[$id] = [102, pack('Na*', strlen($handle), $handle)];
break;
case 4: // CLOSE
if ($this->closeFails) {
return false;
}
$handle = substr($data, 4, unpack('N', $data)[1]);
unset($this->openHandles[$handle]);
$this->answers[$id] = [101, pack('N', 0)];
break;
case 5: // READ
$len = unpack('N', $data)[1];
$handle = substr($data, 4, $len);
['hi' => $hi, 'lo' => $lo, 'n' => $n] = unpack('Nhi/Nlo/Nn', substr($data, 4 + $len));
$offset = $hi * 4294967296 + $lo;
$this->reads[] = $n;
if ($this->throwAt !== null && $this->throwAt >= 0 && $offset >= $this->throwAt) {
throw new \RuntimeException('Connection closed by server');
}
if ($this->loseAt !== null && $this->loseAt >= 0 && $offset >= $this->loseAt) {
$this->answers = [];
return false;
}
$file = $this->files[$this->openHandles[$handle]];
if ($this->failAt !== null && $offset >= $this->failAt) {
$this->answers[$id] = [101, pack('N', 4)]; // FX_FAILURE
} elseif ($offset >= strlen($file)) {
$this->answers[$id] = [101, pack('N', 1)]; // FX_EOF
} else {
$chunk = substr($file, $offset, min($n, $this->cap));
$this->answers[$id] = [103, pack('Na*', strlen($chunk), $chunk)];
}
$this->maxInFlight = max($this->maxInFlight, count($this->answers));
break;
default:
throw new \LogicException("unexpected packet type $type");
}
return true;
}
public function _get_sftp_packet(?int $id = null): string|false {
if ($this->answers === []) {
throw new \LogicException('waiting for an answer that will never come');
}
if ($id === null || !isset($this->answers[$id])) {
throw new \LogicException("no answer for request $id");
}
[$this->packet_type, $payload] = $this->answers[$id];
unset($this->answers[$id]);
return $payload;
}
}
@@ -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');
}
}
+37
View File
@@ -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());
}
}
+172
View File
@@ -0,0 +1,172 @@
<?php
declare(strict_types=1);
namespace OCA\NgBackup\Tests\Unit\Repository;
use OCA\NgBackup\Backend\IBackend;
use OCA\NgBackup\Backend\LocalBackend;
use OCA\NgBackup\Crypto\KeyRing;
use OCA\NgBackup\Repository\Repository;
use OCA\NgBackup\Tests\Unit\TempDirTrait;
use PHPUnit\Framework\TestCase;
/** Restores and deep verifies read a pack as one stream instead of one request per blob. */
class PackReadTest extends TestCase {
use TempDirTrait;
private string $root;
private KeyRing $keys;
private string $src;
private string $snapshot;
protected function setUp(): void {
$this->root = $this->tempDir();
$this->keys = KeyRing::generate();
$this->src = $this->tempDir();
for ($i = 0; $i < 200; $i++) {
file_put_contents(sprintf('%s/f%03d', $this->src, $i), random_bytes(20000));
}
$backend = new LocalBackend($this->root);
$repo = Repository::initWithKey($backend, $this->keys, $this->keys->wrap('pw'));
$this->snapshot = $repo->backupDirectory($this->src)['snapshot'];
}
private function open(PackOpenCountingBackend $backend): Repository {
return Repository::openWithKey($backend, $this->keys, null, null);
}
public function testFullRestoreOpensEachPackOnce(): void {
$backend = new PackOpenCountingBackend(new LocalBackend($this->root));
$target = $this->tempDir();
$this->assertSame(200, $this->open($backend)->restore($this->snapshot, $target));
for ($i = 0; $i < 200; $i++) {
$name = sprintf('f%03d', $i);
$this->assertSame(file_get_contents("$this->src/$name"), file_get_contents("$target/$name"));
}
// One pack: one stream instead of 200 requests.
$this->assertSame(1, $backend->packOpens());
}
public function testDeepVerifyOpensEachPackOnce(): void {
$backend = new PackOpenCountingBackend(new LocalBackend($this->root));
$result = $this->open($backend)->verify($this->snapshot, deep: true);
$this->assertSame(200 * 20000, $result['bytesChecked']);
$this->assertSame(1, $backend->packOpens());
}
public function testReadingBackwardsOpensThePackAgain(): void {
$backend = new PackOpenCountingBackend(new LocalBackend($this->root));
$repo = $this->open($backend);
$this->assertSame(1, $repo->restore($this->snapshot, $this->tempDir(), 'f150'));
$this->assertSame(1, $repo->restore($this->snapshot, $this->tempDir(), 'f050'));
$this->assertSame(2, $backend->packOpens());
}
public function testThePackStreamIsClosedBeforeAnyOtherRequest(): void {
$backend = new PackOpenCountingBackend(new LocalBackend($this->root));
$repo = $this->open($backend);
$this->assertSame(1, $repo->restore($this->snapshot, $this->tempDir(), 'f010'));
$this->assertTrue($backend->packStreamOpen(), 'kept open for the next read');
$repo->putObject('objects/after-restore', 'data');
$this->assertSame([false], $backend->packStreamOpenAtPut);
}
public function testAPackStreamThatCannotSeekIsReadUpToTheOffset(): void {
$backend = new PackOpenCountingBackend(new LocalBackend($this->root), $this->root);
$target = $this->tempDir();
$this->assertSame(1, $this->open($backend)->restore($this->snapshot, $target, 'f150'));
$this->assertSame(file_get_contents("$this->src/f150"), file_get_contents("$target/f150"));
}
public function testDeepVerifyOfATruncatedPackStillChecksTheBlobsBeforeTheDamage(): void {
$packs = (new LocalBackend($this->root))->list('packs');
$this->assertCount(1, $packs);
$pack = $this->root . '/' . $packs[0];
$fh = fopen($pack, 'r+b');
ftruncate($fh, intdiv(filesize($pack), 2));
fclose($fh);
$result = $this->open(new PackOpenCountingBackend(new LocalBackend($this->root)))->verify($this->snapshot, deep: true);
$this->assertGreaterThan(80 * 20000, $result['bytesChecked']);
$this->assertNotSame([], $result['unreadable']);
$this->assertSame([], $result['failed']);
}
}
final class PackOpenCountingBackend implements IBackend {
/** @var list<string> */
public array $gets = [];
/** @var list<bool> */
public array $packStreamOpenAtPut = [];
/** @var resource|null */
private $lastPackStream = null;
/** @param ?string $pipeFrom serve packs as pipes (streams that cannot seek) from this local repository root */
public function __construct(
private IBackend $inner,
private ?string $pipeFrom = null,
) {
}
public function packStreamOpen(): bool {
return is_resource($this->lastPackStream);
}
public function put(string $path, $stream): void {
$this->packStreamOpenAtPut[] = $this->packStreamOpen();
$this->inner->put($path, $stream);
}
public function packOpens(): int {
return count(array_filter($this->gets, fn (string $p): bool => str_starts_with($p, 'packs/')));
}
public function get(string $path) {
$this->gets[] = $path;
if (!str_starts_with($path, 'packs/')) {
return $this->inner->get($path);
}
$fh = $this->pipeFrom !== null
? popen('cat ' . escapeshellarg($this->pipeFrom . '/' . $path), 'r')
: $this->inner->get($path);
$this->lastPackStream = $fh;
return $fh;
}
public function getRange(string $path, int $offset, int $length): string {
$this->gets[] = $path;
return $this->inner->getRange($path, $offset, $length);
}
public function exists(string $path): bool {
return $this->inner->exists($path);
}
public function move(string $from, string $to): void {
$this->inner->move($from, $to);
}
public function list(string $prefix): array {
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);
}
}
+13 -2
View File
@@ -129,7 +129,7 @@ class VerifyTest extends TestCase {
}
}
/** Delegates to a real backend, but getRange() throws for the first $failures calls. */
/** Delegates to a real backend, but reading a pack throws for the first $failures reads. */
final class FlakyRangeBackend implements IBackend {
public function __construct(
private IBackend $inner,
@@ -143,15 +143,22 @@ final class FlakyRangeBackend implements IBackend {
}
public function get(string $path) {
if (str_starts_with($path, 'packs/')) {
$this->fail();
}
return $this->inner->get($path);
}
public function getRange(string $path, int $offset, int $length): string {
$this->fail();
return $this->inner->getRange($path, $offset, $length);
}
private function fail(): void {
if ($this->failures > 0) {
$this->failures--;
throw $this->error ?? new BackendException('connection reset');
}
return $this->inner->getRange($path, $offset, $length);
}
public function exists(string $path): bool {
@@ -166,6 +173,10 @@ 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);
}
+10
View File
@@ -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;
}
}
});