Cluster Configuration for 1C-Bitrix

Cluster Configuration for 1C-Bitrix

A single server cannot scale indefinitely. Under peak loads (promotions, holidays) the site goes down or responds in 10+ seconds — vertical scaling hits cost and physical hardware limits. 1C-Bitrix supports horizontal scaling through its built-in web cluster mechanism: multiple application servers, a shared database with replication, and distributed cache.

1C-Bitrix Cluster Architecture

Standard scheme for high load:

             [Load Balancer]
            /       |        \
     [web-1]    [web-2]    [web-3]
        |           |           |
     [Shared Storage - NFS/GlusterFS]
        |
     [DB Master] ---> [DB Replica-1]
                  ---> [DB Replica-2]
        |
     [Memcached / Redis Cluster]
     [Elasticsearch Cluster]

All web nodes work with a single file storage, shared database, and shared cache. File uploads (images, price lists) go to a shared storage accessible by all nodes.

Requirements for a Clustered Project

Before migrating to a cluster, verify:

• No session data stored in $_SESSION without a shared session storage
• No direct writes to the local filesystem (temp files in /tmp on shared storage, cache in Memcached)
• No hardcoded paths dependent on a specific server
• 1C-Bitrix cache files (/bitrix/cache/) are either NFS-mounted or moved to Memcached

Configuring the Web Cluster Module

In the admin panel: Management → Performance → Cluster.

Activation via PHP:

\Bitrix\Main\Loader::includeModule('cluster');

// Register cluster nodes
$cluster = new \CCluster();
$cluster->Add([
    'NAME' => 'web-02',
    'HOST' => '10.0.0.12',
    'PORT' => 80,
    'STATUS' => 'ACTIVE',
]);

Shared Storage: NFS vs GlusterFS

NFS — simpler to configure, suitable for 2–3 nodes in the same datacenter:

# On the NFS server
apt install nfs-kernel-server
echo "/var/www/bitrix/upload 10.0.0.0/24(rw,sync,no_root_squash)" >> /etc/exports
exportfs -a

# On the web nodes
apt install nfs-common
mount -t nfs 10.0.0.20:/var/www/bitrix/upload /var/www/bitrix/upload

Mount only directories with user-generated content: upload/, cache/ (if not Redis), resize_cache/.

GlusterFS — a distributed filesystem with replication, no single point of failure. More complex to configure, but no SPOF when the NFS server goes down.

Distributed Cache

Without a shared cache, each web node has its own isolated file-based cache. After a product update, invalidation occurs only on one node — the others serve stale data.

// /bitrix/.settings.php — identical on all nodes
'cache' => [
    'value' => [
        'type' => 'memcache',
        'memcache' => [
            ['host' => '10.0.0.30', 'port' => 11211],
            ['host' => '10.0.0.31', 'port' => 11211],
        ],
        'sid' => 'bitrix_production',
    ],
],

Synchronizing Configuration Files

.settings.php, dbconn.php, and php_interface/ must be identical on all nodes. Use rsync via cron or Ansible:

# Master node syncs configs to the others
rsync -az /var/www/bitrix/bitrix/.settings.php web-02:/var/www/bitrix/bitrix/
rsync -az /var/www/bitrix/bitrix/.settings.php web-03:/var/www/bitrix/bitrix/

In production environments, configuration is stored in Git and deployed via CI/CD simultaneously to all nodes.

Timeline

Designing and deploying a 3-node web cluster with NFS storage, database replication, and Memcached takes 5–10 business days, depending on project complexity and the current state of the infrastructure.