Backups to SMB no longer crawl; add occ backup:cancel
Each 30-second step opened the location again and read its list of index files. Over SMB that cost a request per file, so on a large repository opening took longer than the step: each step stored one 4 MiB pack and the list grew further. A first backup of nc01 to a Storage Box over SMB managed 12 GiB in 35 hours. occ backup:run and system cron now keep the repository open for the whole run (reopened when another process took a step or the location was edited), opening no longer counts towards the step's time, and SMB folders are listed in one request. occ backup:cancel stops a running backup: no new step starts and the step in progress finishes first. A run of an earlier version, which does not hold back its next step, is marked cancelled after two minutes unless it is recording its snapshot.
This commit is contained in:
@@ -14,6 +14,7 @@ use OCA\NgBackup\Db\Snapshot;
|
||||
use OCA\NgBackup\Db\SnapshotMapper;
|
||||
use OCA\NgBackup\Db\Target;
|
||||
use OCA\NgBackup\Job\BackupRun;
|
||||
use OCA\NgBackup\Repository\Repository;
|
||||
use OCA\NgBackup\Repository\RepositoryException;
|
||||
use OCP\AppFramework\Db\DoesNotExistException;
|
||||
use OCP\App\IAppManager;
|
||||
@@ -34,6 +35,9 @@ use Psr\Log\LoggerInterface;
|
||||
final class BackupService {
|
||||
public const KIND_FULL = 'full';
|
||||
|
||||
/** @var \WeakMap<Repository, string> the location settings each kept repository was opened with */
|
||||
private \WeakMap $openedWith;
|
||||
|
||||
public function __construct(
|
||||
private TargetService $targets,
|
||||
private KeyService $keys,
|
||||
@@ -47,6 +51,9 @@ final class BackupService {
|
||||
private AlertService $alerts,
|
||||
private LeaseService $leases,
|
||||
) {
|
||||
/** @var \WeakMap<Repository, string> $opened */
|
||||
$opened = new \WeakMap();
|
||||
$this->openedWith = $opened;
|
||||
}
|
||||
|
||||
public function start(Target $target, string $label = ''): Run {
|
||||
@@ -144,13 +151,22 @@ final class BackupService {
|
||||
/**
|
||||
* Do one step of a run until $deadline. Safe to call concurrently: a second caller finds the
|
||||
* run locked and returns immediately.
|
||||
*
|
||||
* @param Repository|null $repo the location's repository, opened by an earlier step of this run
|
||||
* in this process (see runToCompletion()); set when this step opens it
|
||||
*/
|
||||
public function step(Run $run, float $deadline): Run {
|
||||
public function step(Run $run, float $deadline, ?Repository &$repo = null): Run {
|
||||
// Leases expire on their own: a step killed by a time limit or OOM blocks nothing for long.
|
||||
// A step can run well past its deadline (the commit of a large snapshot writes its tree and
|
||||
// index), so the lease holds at least 15 minutes; should it still expire and another process
|
||||
// take the run over, the checks below keep the two from overwriting each other.
|
||||
$ttl = (int)min(3600, max(900, $deadline - microtime(true) + 600));
|
||||
// No new step once a cancel has begun: it waits for the step in progress, if any.
|
||||
try {
|
||||
$this->leases->release($this->leases->acquire(self::cancelKey($run->getId()), LeaseService::SHARED, 60));
|
||||
} catch (LockedException) {
|
||||
return $run;
|
||||
}
|
||||
try {
|
||||
$runLease = $this->leases->acquire('run/' . $run->getId(), LeaseService::EXCLUSIVE, $ttl);
|
||||
} catch (LockedException) {
|
||||
@@ -164,6 +180,7 @@ final class BackupService {
|
||||
return $run;
|
||||
}
|
||||
$loadedAt = null;
|
||||
$savedBefore = $run->getUpdatedAt();
|
||||
try {
|
||||
$run = $this->runs->find($run->getId()); // fresh state after acquiring the lock
|
||||
if ($run->getStatus() !== Run::RUNNING || $run->getKind() !== self::KIND_FULL) {
|
||||
@@ -171,7 +188,19 @@ final class BackupService {
|
||||
}
|
||||
$loadedAt = $run->getUpdatedAt();
|
||||
$target = $this->targets->get((string)$run->getTargetId());
|
||||
$repo = $this->targets->repository($target);
|
||||
// Reopen when another process took a step since ours (its packs are not in our copy of
|
||||
// the index) or the location was edited (password, folder, size limit).
|
||||
$settings = json_encode([$target->getBackend(), $target->getAuth(), $target->getOptions(),
|
||||
$target->getBasePath(), $target->getRepositoryId(), $target->getMaxBytes()], JSON_THROW_ON_ERROR);
|
||||
if ($repo === null || $loadedAt !== $savedBefore || ($this->openedWith[$repo] ?? null) !== $settings) {
|
||||
// Opening reads the index listing, which takes minutes on a slow location with many
|
||||
// packs: that time does not count, or each step would store one blob and stop.
|
||||
$opening = microtime(true);
|
||||
$repo = null;
|
||||
$repo = $this->targets->repository($target);
|
||||
$this->openedWith[$repo] = $settings;
|
||||
$deadline += microtime(true) - $opening;
|
||||
}
|
||||
$state = json_decode($run->getState(), true, 512, JSON_THROW_ON_ERROR);
|
||||
|
||||
if ($run->getPhase() === 'db') {
|
||||
@@ -234,8 +263,14 @@ final class BackupService {
|
||||
$run->setStats(json_encode(['db' => $state['dbStats'] ?? null, 'files' => $state['files']['stats']], JSON_THROW_ON_ERROR));
|
||||
// Another process may have taken the run over after this step's lease expired: then its
|
||||
// writes stand and this step's are dropped.
|
||||
return $this->runs->updateIfUnchanged($run, $loadedAt) ?? $this->runs->find($run->getId());
|
||||
$saved = $this->runs->updateIfUnchanged($run, $loadedAt);
|
||||
if ($saved === null) {
|
||||
$repo = null; // the run is another process's now: its packs are not in our index
|
||||
return $this->runs->find($run->getId());
|
||||
}
|
||||
return $saved;
|
||||
} catch (\Throwable $e) {
|
||||
$repo = null; // may hold blobs of a half-written pack: the next step opens it afresh
|
||||
if ($loadedAt === null) {
|
||||
// The run could not even be loaded: there is nothing fresh to mark as failed, and
|
||||
// the caller's copy may be stale.
|
||||
@@ -266,14 +301,16 @@ final class BackupService {
|
||||
|
||||
/** Run to completion in this process (occ, system cron), reporting progress. */
|
||||
public function runToCompletion(Run $run, ?callable $progress = null, float $stepSeconds = 30): Run {
|
||||
$repo = null; // opened once, by the first step, and kept for the next ones
|
||||
while ($run->getStatus() === Run::RUNNING) {
|
||||
$before = $run->getUpdatedAt();
|
||||
$run = $this->step($run, microtime(true) + $stepSeconds);
|
||||
$run = $this->step($run, microtime(true) + $stepSeconds, $repo);
|
||||
if ($run->getStatus() === Run::RUNNING && $run->getUpdatedAt() === $before) {
|
||||
// Step was not possible (another process holds the run, or a crashed step's lease has
|
||||
// not expired yet): wait instead of spinning.
|
||||
sleep(5);
|
||||
$run = $this->runs->find($run->getId());
|
||||
$repo = null; // another process may have stored packs in the meantime
|
||||
}
|
||||
if ($progress !== null) {
|
||||
$progress($run);
|
||||
@@ -282,11 +319,70 @@ final class BackupService {
|
||||
return $run;
|
||||
}
|
||||
|
||||
public function cancel(Run $run): Run {
|
||||
$run->setStatus(Run::CANCELLED);
|
||||
$run->setFinishedAt(time());
|
||||
$run->setUpdatedAt(time());
|
||||
return $this->runs->update($run);
|
||||
/**
|
||||
* Stop a running backup. First no new step may start, then this waits (up to $wait seconds)
|
||||
* for the step in progress, so a step that is recording the snapshot finishes first. A process
|
||||
* running the backup then finds it cancelled and stops. The packs it stored stay until the
|
||||
* next cleanup.
|
||||
*
|
||||
* A process of an earlier version does not hold back its next step and may never leave a gap:
|
||||
* after $wait the run is marked cancelled while it works, unless it is recording the snapshot.
|
||||
* Its step then fails to save and the process stops.
|
||||
*/
|
||||
public function cancel(Run $run, int $wait = 120): Run {
|
||||
$id = $run->getId();
|
||||
$until = time() + $wait;
|
||||
$hold = null;
|
||||
for ($try = 0; $hold === null; $try++) {
|
||||
try {
|
||||
$hold = $this->leases->acquire(self::cancelKey($id), LeaseService::EXCLUSIVE, $wait + 300);
|
||||
} catch (LockedException) {
|
||||
if ($try >= 20) {
|
||||
throw new \RuntimeException("Run $id is being cancelled already");
|
||||
}
|
||||
usleep(100000); // a step checking whether it may start
|
||||
}
|
||||
}
|
||||
try {
|
||||
$lease = null;
|
||||
while ($lease === null && time() < $until) {
|
||||
try {
|
||||
$lease = $this->leases->acquire('run/' . $id, LeaseService::EXCLUSIVE, 60);
|
||||
} catch (LockedException) {
|
||||
sleep(2);
|
||||
}
|
||||
}
|
||||
try {
|
||||
for ($try = 0; $try < 10; $try++) {
|
||||
$run = $this->runs->find($id);
|
||||
if ($run->getStatus() !== Run::RUNNING) {
|
||||
return $run;
|
||||
}
|
||||
$state = json_decode($run->getState(), true) ?: [];
|
||||
if ($lease === null && ($state['files']['phase'] ?? '') === 'commit') {
|
||||
throw new \RuntimeException("Run $id is recording its snapshot; try again in a minute");
|
||||
}
|
||||
$loadedAt = $run->getUpdatedAt();
|
||||
$run->setStatus(Run::CANCELLED);
|
||||
$run->setFinishedAt(time());
|
||||
if (($cancelled = $this->runs->updateIfUnchanged($run, $loadedAt)) !== null) {
|
||||
return $cancelled;
|
||||
}
|
||||
usleep(200000); // a step of another process saved just now
|
||||
}
|
||||
throw new \RuntimeException("Run $id keeps changing; try again");
|
||||
} finally {
|
||||
if ($lease !== null) {
|
||||
$this->leases->release($lease);
|
||||
}
|
||||
}
|
||||
} finally {
|
||||
$this->leases->release($hold);
|
||||
}
|
||||
}
|
||||
|
||||
private static function cancelKey(int $runId): string {
|
||||
return 'run/' . $runId . '/cancel';
|
||||
}
|
||||
|
||||
private function recordSnapshot(Run $run, array $files): void {
|
||||
|
||||
Reference in New Issue
Block a user