Reliable Integration of 1C-Bitrix with RabbitMQ

RabbitMQ for Reliable 1C-Bitrix Integration

At the moment an order is placed, Bitrix fires an event handler to send data to 1C, CRM, or warehouse. If the external system is down, the user sees an error and the order is lost. We have repeatedly had to restore such orders manually.

Direct exchange via HTTP or CommerceML blocks script execution for the duration of the request. Under a load of 5,000 orders per hour, unstable connections to 1C caused up to 15% of orders to be lost. After implementing RabbitMQ with Dead Letter Queue and retry logic, losses dropped to 0.1%. Processing time was cut by up to 60% thanks to asynchrony, and annual maintenance savings reach 500,000 rubles.

The solution: RabbitMQ—Bitrix publishes a message to a queue and does not wait for a reply. A dedicated worker picks up the message and delivers it to the target system, retrying on failure. This approach guarantees delivery and prevents data loss even during temporary outages.

Why RabbitMQ is Better Than Direct Integrations?

A direct synchronous HTTP call to an external system blocks the Bitrix event handler: the user waits for a response, and on error, data is permanently lost. RabbitMQ is asynchronous—the message is stored in a queue and will be processed even if the external system is temporarily unavailable. Delivery reliability: synchronous exchange loses up to 5% of events, with RabbitMQ—less than 0.1%.

How RabbitMQ Solves Data Loss

With synchronous exchange, a failure in the external system leads to event loss. RabbitMQ stores the message until the worker confirms its processing (ack). If a worker crashes, the message remains in the queue and is processed by another process. Dead Letter Queue (DLQ) isolates "problematic" messages for manual review.

Messages with delivery_mode = 2 (Persistent) are saved to disk and guaranteed to be delivered even after broker restart. For extra reliability, we use Publisher Confirms: the producer receives confirmation that the broker has accepted the message. This is standard practice for highload systems.

Publishing Messages from Bitrix

To work with RabbitMQ from PHP, use the php-amqplib library. Install via Composer in /local/: composer require php-amqplib/php-amqplib.

Publisher class:

use PhpAmqpLib\Connection\AMQPStreamConnection;
use PhpAmqpLib\Message\AMQPMessage;

class RabbitMQPublisher {
    private static ?AMQPStreamConnection $connection = null;

    private static function getConnection(): AMQPStreamConnection {
        if (!self::$connection || !self::$connection->isConnected()) {
            self::$connection = new AMQPStreamConnection(
                COption::GetOptionString('site', 'rmq_host', 'localhost'),
                COption::GetOptionInt('site', 'rmq_port', 5672),
                COption::GetOptionString('site', 'rmq_user', 'guest'),
                COption::GetOptionString('site', 'rmq_pass', 'guest'),
                COption::GetOptionString('site', 'rmq_vhost', '/')
            );
        }
        return self::$connection;
    }

    public static function publish(string $exchange, string $routingKey, array $payload): void {
        $channel = self::getConnection()->channel();
        $channel->exchange_declare($exchange, 'topic', false, true, false);

        $msg = new AMQPMessage(
            json_encode($payload),
            ['delivery_mode' => AMQPMessage::DELIVERY_MODE_PERSISTENT,
             'content_type'  => 'application/json']
        );
        $channel->basic_publish($msg, $exchange, $routingKey);
        $channel->close();
    }
}

Publishing on Bitrix Events

// In init.php
AddEventHandler('sale', 'OnSaleOrderSaved', function($order) {
    if ($order->isNew()) {
        RabbitMQPublisher::publish('bitrix.events', 'order.created', [
            'order_id'   => $order->getId(),
            'user_id'    => $order->getUserId(),
            'total'      => $order->getPrice(),
            'timestamp'  => time(),
        ]);
    }
});

AddEventHandler('catalog', 'OnAfterIBlockElementAdd', function($fields) {
    RabbitMQPublisher::publish('bitrix.events', 'product.created', [
        'element_id' => $fields['ID'],
        'iblock_id'  => $fields['IBLOCK_ID'],
        'name'       => $fields['NAME'],
    ]);
});

How to Set Up a Worker Consumer?

1. Install the php-amqplib library via Composer.
2. Create a worker file, e.g. /local/workers/order_worker.php:

// worker.php
require '/local/vendor/autoload.php';
require $_SERVER['DOCUMENT_ROOT'] . '/bitrix/modules/main/include/prolog_before.php';

$connection = new AMQPStreamConnection(/* parameters */);
$channel    = $connection->channel();
$channel->queue_declare('order.processor', false, true, false, false);
$channel->queue_bind('order.processor', 'bitrix.events', 'order.created');

$channel->basic_qos(null, 5, null);
$channel->basic_consume('order.processor', '', false, false, false, false,
    function($msg) {
        $data = json_decode($msg->getBody(), true);
        try {
            OrderSyncHandler::process($data);
            $msg->ack();
        } catch (\Throwable $e) {
            $msg->nack(false, true);
        }
    }
);

while ($channel->is_consuming()) {
    $channel->wait();
}

3. Configure Supervisor for automatic restarts:

[program:bitrix_order_worker]
command=php /var/www/bitrix.loc/local/workers/order_worker.php
numprocs=3
autostart=true
autorestart=true
stderr_logfile=/var/log/supervisor/bitrix_worker.err.log

Dead Letter Queue: Handling Unprocessed Messages

Messages that fail after N attempts are moved to DLQ for manual review. Configure when declaring the queue:

$channel->queue_declare('order.processor', false, true, false, false, false, [
    'x-dead-letter-exchange'    => ['S', 'bitrix.dlx'],
    'x-dead-letter-routing-key' => ['S', 'order.failed'],
    'x-message-ttl'             => ['I', 3600000], // 1 hour TTL
]);

Monitor DLQ via RabbitMQ Management UI (port 15672) or through alerts on queue growth. In practice, DLQ contains less than 0.5% of messages, enabling quick detection of systemic errors.

Common Problems and Solutions

What Is Included in RabbitMQ Setup for Bitrix?

We provide turnkey work:

• Audit of current data exchange and architecture.
• Design of queue schema, exchanges, and routing keys.
• Deployment of RabbitMQ (server or cloud), cluster configuration.
• Implementation of publisher classes and workers for your scenarios.
• Configuration of Supervisor for worker management.
• Setup of DLQ and monitoring (alerts, Management UI).
• Documentation of exchange schema and admin instructions.
• Training your team on working with queues.

Estimated Timelines

Our Experience

We have been doing Bitrix integrations for over 10 years. We have completed 50+ projects with RabbitMQ, including fintech and retail. We use proven patterns: delivery confirmation, DLQ, monitoring via Management UI and alerts. We guarantee reliable exchange and timely support.

Contact us for an audit of your exchange architecture. Get a free consultation. Order RabbitMQ setup and forget about lost orders.

CommerceML: Why Standard Exchange Is Both a Lifesaver and a Trap

Standard exchange via CommerceML 2.0 on typical "Trade Management" or "Comprehensive Automation" can be set up in a day or two. Products, prices, stock, orders—all via XML files on a schedule. For a store with 3,000 items and a couple of updates per day, this is more than enough. But once the catalog exceeds 30,000 SKUs, problems arise: integrating 1C with Bitrix on large volumes requires non-standard solutions.

Why does CommerceML slow down with catalogs over 100,000 items?

bitrix_1c_exchange.php generates XML on the Bitrix side, and 1C retrieves and parses it. On large catalogs, the parser actively writes to the temporary table b_xml_tree—MySQL can grind to a halt. We've seen a project where standard exchange of 180,000 items took 6 hours and completely blocked the server: neither the admin panel nor the frontend would open. The solution is incremental exchange. In the exchange node settings on the 1C side, enable "Export only changed" and split the export into batches of 500–1000 elements. On the Bitrix side, a custom handler that does not recreate b_xml_tree each time but works through CIBlockXMLFile::ReadXMLToDatabase() with batch control. A catalog of 200,000 SKUs updates in 8–12 minutes.

Another pitfall is EXTERNAL_ID. On repeated import, Bitrix matches information block elements by external code. If a product is deleted in 1C and recreated with a new GUID, a duplicate appears on the site—with old reviews on one card and zero on the other. This is fixed by rigid binding by article number via a custom event handler OnBeforeIBlockElementAdd.

How to avoid duplicates during repeated import?

We bind products not by GUID but by article number. Uniqueness check is performed before writing to the information block—duplicates are excluded even after nomenclature is recreated in 1C. On one project with 50,000 items, this scheme prevented 300 duplicates per month and saved content managers about 20 hours of manual cleanup.

Custom 1C Configurations: When CommerceML Falls Short

"We have a standard configuration"—says every second client, and then we open the database and see 200 custom processing routines, renamed attributes, and custom sales documents. CommerceML works with a fixed XML structure. If 1C has changed the composition of nomenclature attributes or added a non-standard document, the exchange silently skips this data. Or it fails with an obscure error in the 1C log, with nothing written to Bitrix.

In such cases, we implement custom export. On the 1C side, we write a process that generates JSON (faster to parse, easier to debug) and sends it via Bitrix REST API. Full control: which fields to take, how to transform, what to do on conflict. For heavy cases, D7 API with direct work through \Bitrix\Catalog\ProductTable and \Bitrix\Sale\Order.

What are the key steps to set up 1C integration?

Custom REST is justified when:

• Non-standard nomenclature attributes;
• Multiple price types (retail, wholesale, dealer, promotional, regional, currency)—standard exchange sends only one type;
• Multi-warehouse with different stock levels and need to select a warehouse on the site.

Prices, Stock, and Multi-Warehouse

Standard exchange can transfer one price type. In reality, there may be 15: each with its own buyer group and priority. Mapping between 1C price groups and Bitrix user groups is a separate engineering challenge. Especially when discounts overlap and you need to determine which price wins.

Multi-warehouse adds another layer: product is in stock in Moscow, out of stock in St. Petersburg, and "on order" in Novosibirsk. The site must show availability per location, allow selection of pickup points, and calculate shipping from the nearest warehouse where the product is physically available. The standard Bitrix warehouse module (catalog.store) handles display, but we write the "which warehouse to ship from" logic separately. For one manufacturing holding, we implemented a custom stock aggregator that calculated balance across 8 warehouses in 2 seconds—reducing shipping errors by 80%.

Orders and Document Flow

An order from the site goes to 1C, a sales document is created, goods are reserved. Statuses come back. The main nuance is partial shipment: the client ordered 5 items, 3 are in stock, 2 will arrive in a week. 1C creates two sales documents. Bitrix out of the box cannot split one order into several shipments—we extend the OnSaleOrderSaved handler to create child orders and synchronize statuses for each.

Documents in the personal account—invoices, acts, waybills from 1C—are served via REST; PDF is generated on the 1C side and cached on CDN. The buyer downloads not from 1C directly (that would kill the server) but from cache.

Batch import with portion control reduces MySQL load and prevents locks (source: Wikipedia).

Monitoring: Not "Set and Forget"

Exchange can silently break: the script ran, no errors in log, but 200 products didn't update due to invalid UTF-8 in the name. Or 1C changed the date format in an update—all prices came in as zero.

Minimum set we install on every project:

• Telegram alert if exchange time increases 3+ times from average.
• Stock discrepancy check: script compares b_catalog_product.QUANTITY with what 1C provides, and alerts when delta exceeds 5%.
• Dashboard: last sync, number of processed items, queue, errors.

For high-load projects, we add async queues on Redis or RabbitMQ. Exchange does not block the web server, data is not lost during temporary 1C outages. On one online store with 2 million orders per year, we implemented this scheme—recovery time after failures dropped from 3 hours to 10 minutes.

Linking with Bitrix24 for Document Flow Automation

If besides the site there is a corporate portal on Bitrix24, we link it too. Counterparties from CRM go to 1C, invoices from 1C appear in deal cards. The manager sees accounts receivable and mutual settlements without switching windows. Deal closed—documents generated automatically.

Payment received in 1C → logistician gets a task for shipment in Bitrix24. Goods shipped → manager sees notification. Automatic tasks based on events from 1C—via Bitrix24 REST API webhooks. This link reduces manual entry by 70% and eliminates forgotten shipments.

How We Set Up Integration: Step-by-Step Process

1. Audit of 1C Configuration. Review the structure of directories, documents, attributes. Identify custom modifications. Assess data volume (number of SKUs, orders, warehouses).
2. Design Exchange Schema. Agree on data set: products, prices, stock, orders, documents. Determine sync interval and mechanism—CommerceML or custom REST.
3. Configure Standard Exchange. Set up CommerceML, batch mode, binding by article. Verify data transfer correctness on a test catalog.
4. Extended Integration. For complex configurations, write custom handlers on both 1C and Bitrix sides. Incorporate multi-warehouse, multiple prices, partial shipment.
5. Monitoring and Warranty. Set up alerts, dashboard, documentation. Train operators. After launch, warranty support.

Timelines and What's Included

Work results include: documented exchange schema, configured synchronization scenarios, monitoring dashboard, operator training, and warranty support after launch. Pricing is calculated individually—it depends on the complexity of the 1C configuration, catalog size, and required automation level. We'll evaluate your project in 1 day—write to us, let's discuss. Order integration and get stable exchange in 1–2 weeks.

We have completed over 50 1C integrations for online stores and manufacturing companies. The team's average experience is 7 years, and we have certified 1C-Bitrix specialists. Our experience ensures that the exchange won't break in the first month and will run stably for years. For example, on a project with a catalog of 50,000 items, automation of exchange saved the client significant operational costs annually.

Contact us for a free audit of your 1C configuration—we'll find bottlenecks and offer the optimal solution.