HCL Commerce Search limitations and restrictions

Some features of the Commerce search environment are unsupported for the initial release of the Commerce Search. These features will be restored in future point releases. Some features are deprecated because they have been or will be replaced by specific Ingest or Elasticsearch methods and interfaces.

Planned for future releases

The following search features are still available through the Solr search option, but are unavailable when using the Commerce Search service. They will become available with Ingest in future releases.

  • Language fallback is supported under Commerce Search. If catalog text is not defined in a certain language, the object containing the missing text instead uses the store's default language as a fallback. This assumes that the text is available in the default locale. If the store's default language does not contain the text, the object either is not returned, or is returned with missing properties.
  • Product variant searches are supported with Elasticsearch from Commerce Search. Variants provide an intermediate class of object between product categories and SKUs. For more information, see Product variants.
  • The Oracle database is supported under Commerce Search.
  • The Solr result grouping feature, also known as field collapsing, is implemented in Commerce Search. For more information about how result grouping is used in Commerce+, see Hero product image.
The following features are not yet available for the Elasticsearch-based Search solution:
  • Site content web crawler
  • Unstructured content (attachment) search
  • A landing page specified through Search Term Association cannot be returned via the Query service
  • Support for classic attributes

Limitations

  • Deep Search Sequencing can be done with Elasticsearch, however the sequence is based on the full master catalog as represented in the Commerce Lab, and it performs global sequencing only, which is equivalent to the behavior of Solr with the wc-component.xml configuration parameter EnableDeepProductSequence set to false. For more information about the Solr implementation of Deep Search Sequencing, see Sorting products using deep search sequencing.
  • Deep sequencing is improved to include both category and product sequencing. For more information, see Enabling deep search sequencing.
  • The Search engine does not reconstruct broken words. For example, a search for "chairs" will succeed, as will a search for "chiars." The parser recognizes the string as a single word and performs operations such as spellcheck against that word. However, a search for "cha irs" does not work, because the parser takes this string to be two words, neither of which autocorrects into a meaningful noun.
  • Keywords such as "more than" or "less than" are only turned into numeric operators in the price range filter. A search for "less than two years" does not return a list of products made in the past two years. A search for "less than $20" does return a list of products or services whose prices are below $20.
  • Full reindexing for all indices of the extended (e-site) store is required to reflect on the storefront linked products (from catalog asset store) for a new sales category created in the extended site (e-site) store by assigning products from the catalog asset store.
  • Only one dataload task can be performed at a time. This is because there is no multi-thread related information provided by Dataload in the TI_DELTA_CATENTRY and TI_DELTA_CATGROUP tables. NiFi uses these tables too process incremental updates for dataload.
  • Measurement/dimension matchmaker limitations:
    • For each additional language you use with the matchmaker, raise the NiFi mem_reservation and mem_limit by two gigabytes each.
    • Supports the usage of characters X or x as the dimension separator.
    • Supports only characters L, W, H as dimensions while defining dimensions for a product specification in the catalog.
    • Supports only one-dimensional, two-dimensional, and three-dimensional product searches.
    • Supports dimensional search with numeric value along with dimension keyword, such as long, wide, high. For example, 0.5m long sofa.
    • Space must be provided in between the unit of measure, dimension, and number.
  • If you change the value of a product's descriptive searchable attribute from true to false, the product will remain searchable in the storefront until you perform a full re-index.
  • The Elasticsearch-based search solution does not support using the Classic Attribute model. If your site is still using the Classic Attribute model, migrate your attributes to the Attribute Dictionary model. For additional Attribute Dictionary migration information, see Migrating attribute data to the attribute dictionary in the WebSphere Commerce documentation.
  • The Elasticsearch-based search solution does not respect the Show facets settings in Commerce Lab for the Brand and the Price search facets. Even if you attempt to disable these facets, their values will still be returned by the storefront service.
  • Tier pricing cannot currently be indexed with the Elasticsearch-based Search solution. The indexed contract and offer price is only based on a quantity unit of 1.
  • While searching for a product part number avoid adding other keywords such as category name or brand name. Doing this can return an unexpected result when searching for part numbers.
  • Oracle Database Versions 12c and up have a limit on the Program Global Area (PGA) memory area used by single Oracle server processes. This limit of 2GB can cause an error in the auth.product pipeline. For more information on the error conditions and how to avoid them, see Troubleshooting: auth.product pipeline causes Oracle PGA limit error.
  • While using DB2 databases only, when displaying a product with a large (for example, 50K character) Long Description, the Description content is truncated to a length of 32K characters.

  • Keyword suggestions from the Elasticsearch-based Search API are generated organically by Elasticsearch without any contract-based entitlement filtering. As a result, a suggested keyword can lead to an empty search result.

Discontinued features

These features are no longer available in Commerce Search.

  • You can no longer use the customization programming model of BOD based search on the Transaction server. This service was maintained for backwards compatibility with previous Commerce versions. You can migrate most BOD-based customizations to the Commerce+ model to obviate this problem. For more information, see Migrating BOD-based Search.
  • Price modes (computed, mixed) are not supported. This is a pricing mode where prices are determined at checkout-time. You cannot use these price modes to use the price adjustments from price rules. The price mode that is supported is the indexed price mode wherein all prices (list, offer, contract) are retrieved from the search index. The Ingest service copies into the product index all these prices from the price index. Once the price rule is assigned to the contract, you need to manually build the contract price index so that Ingest service can copy this into the product index and reflect the prices appropriately at the storefront. List and offer prices are indexed directly from the database tables. For more information on price rules, see Price rules: An overview.
  • The Commerce+ search statistics feature applies only to searches made through the Solr search engine. This tool is replaced by Google Analytics starting in Commerce Search. For more information about the Solr statistics management tools, see Managing site search statistics.
  • The Elasticsearch-based search solution no longer includes a built-in default web crawler implementation.
Note: The Elasticsearch-based Search solution no longer includes a default web crawler implementation out of the box.