Important

You are browsing documentation for version 5.1 of OroCommerce, supported until March 2026. Read the documentation for version 6.0 (the latest LTS version) to get up-to-date information.

See our Release Process documentation for more information on the currently supported and upcoming releases.

Optimize Website Indexation and Price Recalculation 

While website indexation and price recalculation background processes are vital for OroCommerce store, they may be time- and resource- consuming.

This topic summarizes best practices and advice that helps you maximize the efficiency of these processes by tuning your software and hardware.

Note

Price recalculation is required before the indexation.

Price Recalculation 

The price list data may contain quite complex dependencies and formula. When price recalculation happens, all of the price list data is analyzed, recalculated if necessary, and saved to the simplified combine tables. This cache-like approach enables quicker data retrieval and increases the OroCommerce storefront responsiveness.

To launch the price recalculation process, run the following console command:

php bin/console oro:price-lists:schedule-recalculate --all

Execution time may vary from seconds on bare factory data to multiple hours for several hundred thousand product prices.

Indexation 

For faster search and access, the business data in the database should be indexed via an internal search engine. After indexation, the user gets prompt search and auto-complete results even for the extremely large amount of raw data. The engine indexes the information (prices, product descriptions, SEO, attributes etc.) and guarantees quick data retrieval leaving the original data untouched.

OroCommerce search index may be persisted in the database layer (using ORM) or on the file system using the ElasticSearch (the latter is available in the Enterprise edition only).

To launch the indexation process, use the following console command:

php bin/console oro:website:reindex

Execution time may range from few minutes to several hours, depending on the hardware and software configuration and the volume of data that is indexed.

To speed up the indexation, you can:

  • Use parallel (multi-thread) indexation. As a prerequisite, ensure that several MessageQueue consumers are running. Next, run the command above with the --scheduled parameter. Depending on the number of CPU cores and consumers running, this can drastically cut the indexation time.

  • Use the --scheduled (multithreaded) indexation to control the amount of entities, indexed per message. This will control indexation load distribution among several consumers and allow scaling. Use an example value of */1000 to generate messages with 1000 products each.

Related Articles