The Lock Component

The Lock Component creates and manages locks, a mechanism to provide exclusive access to a shared resource.

Installation

$ composer require symfony/lock

注釈

If you install this component outside of a Symfony application, you must require the vendor/autoload.php file in your code to enable the class autoloading mechanism provided by Composer. Read this article for more details.

Usage

Locks are used to guarantee exclusive access to some shared resource. In Symfony applications, you can use locks for example to ensure that a command is not executed more than once at the same time (on the same or different servers).

Locks are created using a LockFactory class, which in turn requires another class to manage the storage of locks:

use Symfony\Component\Lock\LockFactory;
use Symfony\Component\Lock\Store\SemaphoreStore;

$store = new SemaphoreStore();
$factory = new LockFactory($store);

The lock is created by calling the createLock() method. Its first argument is an arbitrary string that represents the locked resource. Then, a call to the acquire() method will try to acquire the lock:

// ...
$lock = $factory->createLock('pdf-invoice-generation');

if ($lock->acquire()) {
    // The resource "pdf-invoice-generation" is locked.
    // You can compute and generate invoice safely here.

    $lock->release();
}

If the lock can not be acquired, the method returns false. The acquire() method can be safely called repeatedly, even if the lock is already acquired.

注釈

Unlike other implementations, the Lock Component distinguishes locks instances even when they are created for the same resource. If a lock has to be used by several services, they should share the same Lock instance returned by the LockFactory::createLock method.

ちなみに

If you don’t release the lock explicitly, it will be released automatically on instance destruction. In some cases, it can be useful to lock a resource across several requests. To disable the automatic release behavior, set the third argument of the createLock() method to false.

Blocking Locks

By default, when a lock cannot be acquired, the acquire method returns false immediately. To wait (indefinitely) until the lock can be created, pass true as the argument of the acquire() method. This is called a blocking lock because the execution of your application stops until the lock is acquired.

Some of the built-in Store classes support this feature. When they don’t, they can be decorated with the RetryTillSaveStore class:

use Symfony\Component\Lock\LockFactory;
use Symfony\Component\Lock\Store\RedisStore;
use Symfony\Component\Lock\Store\RetryTillSaveStore;

$store = new RedisStore(new \Predis\Client('tcp://localhost:6379'));
$store = new RetryTillSaveStore($store);
$factory = new LockFactory($store);

$lock = $factory->createLock('notification-flush');
$lock->acquire(true);

Expiring Locks

Locks created remotely are difficult to manage because there is no way for the remote Store to know if the locker process is still alive. Due to bugs, fatal errors or segmentation faults, it cannot be guaranteed that release() method will be called, which would cause the resource to be locked infinitely.

The best solution in those cases is to create expiring locks, which are released automatically after some amount of time has passed (called TTL for Time To Live). This time, in seconds, is configured as the second argument of the createLock() method. If needed, these locks can also be released early with the release() method.

The trickiest part when working with expiring locks is choosing the right TTL. If it’s too short, other processes could acquire the lock before finishing the job; if it’s too long and the process crashes before calling the release() method, the resource will stay locked until the timeout:

// ...
// create an expiring lock that lasts 30 seconds
$lock = $factory->createLock('charts-generation', 30);

$lock->acquire();
try {
    // perform a job during less than 30 seconds
} finally {
    $lock->release();
}

ちなみに

To avoid letting the lock in a locking state, it’s recommended to wrap the job in a try/catch/finally block to always try to release the expiring lock.

In case of long-running tasks, it’s better to start with a not too long TTL and then use the refresh() method to reset the TTL to its original value:

// ...
$lock = $factory->createLock('charts-generation', 30);

$lock->acquire();
try {
    while (!$finished) {
        // perform a small part of the job.

        // renew the lock for 30 more seconds.
        $lock->refresh();
    }
} finally {
    $lock->release();
}

ちなみに

Another useful technique for long-running tasks is to pass a custom TTL as an argument of the refresh() method to change the default lock TTL:

$lock = $factory->createLock('charts-generation', 30);
// ...
// refresh the lock for 30 seconds
$lock->refresh();
// ...
// refresh the lock for 600 seconds (next refresh() call will be 30 seconds again)
$lock->refresh(600);

This component also provides two useful methods related to expiring locks: getExpiringDate() (which returns null or a \DateTimeImmutable object) and isExpired() (which returns a boolean).

The Owner of The Lock

Locks that are acquired for the first time are owned by the Lock instance that acquired it. If you need to check whether the current Lock instance is (still) the owner of a lock, you can use the isAcquired() method:

if ($lock->isAcquired()) {
    // We (still) own the lock
}

Because of the fact that some lock stores have expiring locks (as seen and explained above), it is possible for an instance to lose the lock it acquired automatically:

// If we cannot acquire ourselves, it means some other process is already working on it
if (!$lock->acquire()) {
    return;
}

$this->beginTransaction();

// Perform a very long process that might exceed TTL of the lock

if ($lock->isAcquired()) {
    // Still all good, no other instance has acquired the lock in the meantime, we're safe
    $this->commit();
} else {
    // Bummer! Our lock has apparently exceeded TTL and another process has started in
    // the meantime so it's not safe for us to commit.
    $this->rollback();
    throw new \Exception('Process failed');
}

ご用心

A common pitfall might be to use the isAcquired() method to check if a lock has already been acquired by any process. As you can see in this example you have to use acquire() for this. The isAcquired() method is used to check if the lock has been acquired by the current process only!

[1]	Technically, the true owners of the lock are the ones that share the same instance of Key, not Lock. But from a user perspective, Key is internal and you will likely only be working with the Lock instance so it’s easier to think of the Lock instance as being the one that is the owner of the lock.

Available Stores

Locks are created and managed in Stores, which are classes that implement PersistingStoreInterface and, optionally, BlockingStoreInterface.

The component includes the following built-in store types:

Store	Scope	Blocking	Expiring
FlockStore	local	yes	no
MemcachedStore	remote	no	yes
MongoDbStore	remote	no	yes
PdoStore	remote	no	yes
RedisStore	remote	no	yes
SemaphoreStore	local	yes	no
ZookeeperStore	remote	no	no

FlockStore

The FlockStore uses the file system on the local computer to create the locks. It does not support expiration, but the lock is automatically released when the PHP process is terminated:

use Symfony\Component\Lock\Store\FlockStore;

// the argument is the path of the directory where the locks are created
// if none is given, sys_get_temp_dir() is used internally.
$store = new FlockStore('/var/stores');

ご用心

Beware that some file systems (such as some types of NFS) do not support locking. In those cases, it’s better to use a directory on a local disk drive or a remote store based on PDO, Redis or Memcached.

MemcachedStore

The MemcachedStore saves locks on a Memcached server, it requires a Memcached connection implementing the \Memcached class. This store does not support blocking, and expects a TTL to avoid stalled locks:

use Symfony\Component\Lock\Store\MemcachedStore;

$memcached = new \Memcached();
$memcached->addServer('localhost', 11211);

$store = new MemcachedStore($memcached);

注釈

Memcached does not support TTL lower than 1 second.

MongoDbStore

バージョン 5.1 で追加: The MongoDbStore was introduced in Symfony 5.1.

The MongoDbStore saves locks on a MongoDB server >=2.2, it requires a \MongoDB\Collection or \MongoDB\Client from mongodb/mongodb or a MongoDB Connection String. This store does not support blocking and expects a TTL to avoid stalled locks:

use Symfony\Component\Lock\Store\MongoDbStore;

$mongo = 'mongodb://localhost/database?collection=lock';
$options = [
    'gcProbablity' => 0.001,
    'database' => 'myapp',
    'collection' => 'lock',
    'uriOptions' => [],
    'driverOptions' => [],
];
$store = new MongoDbStore($mongo, $options);

The MongoDbStore takes the following $options (depending on the first parameter type):

Option	Description
gcProbablity	Should a TTL Index be created expressed as a probability from 0.0 to 1.0 (Defaults to 0.001)
database	The name of the database
collection	The name of the collection
uriOptions	Array of uri options for MongoDBClient::__construct
driverOptions	Array of driver options for MongoDBClient::__construct

When the first parameter is a:

MongoDB\Collection:

• $options['database'] is ignored
• $options['collection'] is ignored

MongoDB\Client:

• $options['database'] is mandatory
• $options['collection'] is mandatory

MongoDB Connection String:

• $options['database'] is used otherwise /path from the DSN, at least one is mandatory
• $options['collection'] is used otherwise ?collection= from the DSN, at least one is mandatory

注釈

The collection querystring parameter is not part of the MongoDB Connection String definition. It is used to allow constructing a MongoDbStore using a Data Source Name (DSN) without $options.

PdoStore

The PdoStore saves locks in an SQL database. It requires a PDO connection, a Doctrine DBAL Connection, or a Data Source Name (DSN). This store does not support blocking, and expects a TTL to avoid stalled locks:

use Symfony\Component\Lock\Store\PdoStore;

// a PDO, a Doctrine DBAL connection or DSN for lazy connecting through PDO
$databaseConnectionOrDSN = 'mysql:host=127.0.0.1;dbname=lock';
$store = new PdoStore($databaseConnectionOrDSN, ['db_username' => 'myuser', 'db_password' => 'mypassword']);

注釈

This store does not support TTL lower than 1 second.

Before storing locks in the database, you must create the table that stores the information. The store provides a method called createTable() to set up this table for you according to the database engine used:

try {
    $store->createTable();
} catch (\PDOException $exception) {
    // the table could not be created for some reason
}

A great way to set up the table in production is to call the createTable() method in your local computer and then generate a database migration:

$ php bin/console doctrine:migrations:diff
$ php bin/console doctrine:migrations:migrate

RedisStore

The RedisStore saves locks on a Redis server, it requires a Redis connection implementing the \Redis, \RedisArray, \RedisCluster or \Predis classes. This store does not support blocking, and expects a TTL to avoid stalled locks:

use Symfony\Component\Lock\Store\RedisStore;

$redis = new \Redis();
$redis->connect('localhost');

$store = new RedisStore($redis);

SemaphoreStore

The SemaphoreStore uses the PHP semaphore functions to create the locks:

use Symfony\Component\Lock\Store\SemaphoreStore;

$store = new SemaphoreStore();

CombinedStore

The CombinedStore is designed for High Availability applications because it manages several stores in sync (for example, several Redis servers). When a lock is being acquired, it forwards the call to all the managed stores, and it collects their responses. If a simple majority of stores have acquired the lock, then the lock is considered as acquired; otherwise as not acquired:

use Symfony\Component\Lock\Store\CombinedStore;
use Symfony\Component\Lock\Store\RedisStore;
use Symfony\Component\Lock\Strategy\ConsensusStrategy;

$stores = [];
foreach (['server1', 'server2', 'server3'] as $server) {
    $redis = new \Redis();
    $redis->connect($server);

    $stores[] = new RedisStore($redis);
}

$store = new CombinedStore($stores, new ConsensusStrategy());

Instead of the simple majority strategy (ConsensusStrategy) an UnanimousStrategy can be used to require the lock to be acquired in all the stores.

ご用心

In order to get high availability when using the ConsensusStrategy, the minimum cluster size must be three servers. This allows the cluster to keep working when a single server fails (because this strategy requires that the lock is acquired in more than half of the servers).

ZookeeperStore

The ZookeeperStore saves locks on a ZooKeeper server. It requires a ZooKeeper connection implementing the \Zookeeper class. This store does not support blocking and expiration but the lock is automatically released when the PHP process is terminated:

use Symfony\Component\Lock\Store\ZookeeperStore;

$zookeeper = new \Zookeeper('localhost:2181');
// use the following to define a high-availability cluster:
// $zookeeper = new \Zookeeper('localhost1:2181,localhost2:2181,localhost3:2181');

$store = new ZookeeperStore($zookeeper);

注釈

Zookeeper does not require a TTL as the nodes used for locking are ephemeral and die when the PHP process is terminated.

Reliability

The component guarantees that the same resource can’t be lock twice as long as the component is used in the following way.

Remote Stores

Remote stores (MemcachedStore, MongoDbStore, PdoStore, RedisStore and ZookeeperStore) use a unique token to recognize the true owner of the lock. This token is stored in the Key object and is used internally by the Lock, therefore this key must not be shared between processes (session, caching, fork, ...).

ご用心

Do not share a key between processes.

Every concurrent process must store the Lock in the same server. Otherwise two different machines may allow two different processes to acquire the same Lock.

ご用心

To guarantee that the same server will always be safe, do not use Memcached behind a LoadBalancer, a cluster or round-robin DNS. Even if the main server is down, the calls must not be forwarded to a backup or failover server.

Expiring Stores

Expiring stores (MemcachedStore, MongoDbStore, PdoStore and RedisStore) guarantee that the lock is acquired only for the defined duration of time. If the task takes longer to be accomplished, then the lock can be released by the store and acquired by someone else.

The Lock provides several methods to check its health. The isExpired() method checks whether or not its lifetime is over and the getRemainingLifetime() method returns its time to live in seconds.

Using the above methods, a more robust code would be:

// ...
$lock = $factory->createLock('invoice-publication', 30);

$lock->acquire();
while (!$finished) {
    if ($lock->getRemainingLifetime() <= 5) {
        if ($lock->isExpired()) {
            // lock was lost, perform a rollback or send a notification
            throw new \RuntimeException('Lock lost during the overall process');
        }

        $lock->refresh();
    }

    // Perform the task whose duration MUST be less than 5 minutes
}

ご用心

Choose wisely the lifetime of the Lock and check whether its remaining time to leave is enough to perform the task.

ご用心

Storing a Lock usually takes a few milliseconds, but network conditions may increase that time a lot (up to a few seconds). Take that into account when choosing the right TTL.

By design, locks are stored in servers with a defined lifetime. If the date or time of the machine changes, a lock could be released sooner than expected.

ご用心

To guarantee that date won’t change, the NTP service should be disabled and the date should be updated when the service is stopped.

FlockStore

By using the file system, this Store is reliable as long as concurrent processes use the same physical directory to stores locks.

Processes must run on the same machine, virtual machine or container. Be careful when updating a Kubernetes or Swarm service because for a short period of time, there can be two running containers in parallel.

The absolute path to the directory must remain the same. Be careful of symlinks that could change at anytime: Capistrano and blue/green deployment often use that trick. Be careful when the path to that directory changes between two deployments.

Some file systems (such as some types of NFS) do not support locking.

ご用心

All concurrent processes must use the same physical file system by running on the same machine and using the same absolute path to locks directory.

By definition, usage of FlockStore in an HTTP context is incompatible with multiple front servers, unless to ensure that the same resource will always be locked on the same machine or to use a well configured shared file system.

Files on the file system can be removed during a maintenance operation. For instance, to clean up the /tmp directory or after a reboot of the machine when a directory uses tmpfs. It’s not an issue if the lock is released when the process ended, but it is in case of Lock reused between requests.

ご用心

Do not store locks on a volatile file system if they have to be reused in several requests.

MemcachedStore

The way Memcached works is to store items in memory. That means that by using the MemcachedStore the locks are not persisted and may disappear by mistake at anytime.

If the Memcached service or the machine hosting it restarts, every lock would be lost without notifying the running processes.

ご用心

To avoid that someone else acquires a lock after a restart, it’s recommended to delay service start and wait at least as long as the longest lock TTL.

By default Memcached uses a LRU mechanism to remove old entries when the service needs space to add new items.

ご用心

The number of items stored in Memcached must be under control. If it’s not possible, LRU should be disabled and Lock should be stored in a dedicated Memcached service away from Cache.

When the Memcached service is shared and used for multiple usage, Locks could be removed by mistake. For instance some implementation of the PSR-6 clear() method uses the Memcached’s flush() method which purges and removes everything.

ご用心

The method flush() must not be called, or locks should be stored in a dedicated Memcached service away from Cache.

MongoDbStore

ご用心

The locked resource name is indexed in the _id field of the lock collection. Beware that in MongoDB an indexed field’s value can be a maximum of 1024 bytes in length inclusive of structural overhead.

A TTL index must be used to automatically clean up expired locks. Such an index can be created manually:

db.lock.ensureIndex(
    { "expires_at": 1 },
    { "expireAfterSeconds": 0 }
)

Alternatively, the method MongoDbStore::createTtlIndex(int $expireAfterSeconds = 0) can be called once to create the TTL index during database setup. Read more about Expire Data from Collections by Setting TTL in MongoDB.

ちなみに

MongoDbStore will attempt to automatically create a TTL index. It’s recommended to set constructor option gcProbablity = 0.0 to disable this behavior if you have manually dealt with TTL index creation.

ご用心

This store relies on all PHP application and database nodes to have synchronized clocks for lock expiry to occur at the correct time. To ensure locks don’t expire prematurely; the lock TTL should be set with enough extra time in expireAfterSeconds to account for any clock drift between nodes.

writeConcern, readConcern and readPreference are not specified by MongoDbStore meaning the collection’s settings will take effect. Read more about Replica Set Read and Write Semantics in MongoDB.

PdoStore

The PdoStore relies on the ACID properties of the SQL engine.

ご用心

In a cluster configured with multiple primaries, ensure writes are synchronously propagated to every nodes, or always use the same node.

ご用心

Some SQL engines like MySQL allow to disable the unique constraint check. Ensure that this is not the case SET unique_checks=1;.

In order to purge old locks, this store uses a current datetime to define an expiration date reference. This mechanism relies on all server nodes to have synchronized clocks.

ご用心

To ensure locks don’t expire prematurely; the TTLs should be set with enough extra time to account for any clock drift between nodes.

RedisStore

The way Redis works is to store items in memory. That means that by using the RedisStore the locks are not persisted and may disappear by mistake at anytime.

If the Redis service or the machine hosting it restarts, every locks would be lost without notifying the running processes.

ご用心

To avoid that someone else acquires a lock after a restart, it’s recommended to delay service start and wait at least as long as the longest lock TTL.

ちなみに

Redis can be configured to persist items on disk, but this option would slow down writes on the service. This could go against other uses of the server.

When the Redis service is shared and used for multiple usages, locks could be removed by mistake.

ご用心

The command FLUSHDB must not be called, or locks should be stored in a dedicated Redis service away from Cache.

CombinedStore

Combined stores allow to store locks across several backends. It’s a common mistake to think that the lock mechanism will be more reliable. This is wrong The CombinedStore will be, at best, as reliable as the least reliable of all managed stores. As soon as one managed store returns erroneous information, the CombinedStore won’t be reliable.

ご用心

All concurrent processes must use the same configuration, with the same amount of managed stored and the same endpoint.

ちなみに

Instead of using a cluster of Redis or Memcached servers, it’s better to use a CombinedStore with a single server per managed store.

SemaphoreStore

Semaphores are handled by the Kernel level. In order to be reliable, processes must run on the same machine, virtual machine or container. Be careful when updating a Kubernetes or Swarm service because for a short period of time, there can be two running containers in parallel.

ご用心

All concurrent processes must use the same machine. Before starting a concurrent process on a new machine, check that other process are stopped on the old one.

ご用心

When running on systemd with non-system user and option RemoveIPC=yes (default value), locks are deleted by systemd when that user logs out. Check that process is run with a system user (UID <= SYS_UID_MAX) with SYS_UID_MAX defined in /etc/login.defs, or set the option RemoveIPC=off in /etc/systemd/logind.conf.

ZookeeperStore

The way ZookeeperStore works is by maintaining locks as ephemeral nodes on the server. That means that by using ZookeeperStore the locks will be automatically released at the end of the session in case the client cannot unlock for any reason.

If the ZooKeeper service or the machine hosting it restarts, every lock would be lost without notifying the running processes.

ちなみに

To use ZooKeeper’s high-availability feature, you can setup a cluster of multiple servers so that in case one of the server goes down, the majority will still be up and serving the requests. All the available servers in the cluster will see the same state.

注釈

As this store does not support multi-level node locks, since the clean up of intermediate nodes becomes an overhead, all locks are maintained at the root level.

Overall

Changing the configuration of stores should be done very carefully. For instance, during the deployment of a new version. Processes with new configuration must not be started while old processes with old configuration are still running.