Skip to main content

Index Diagnostic

Index Diagnostic is a modal window for diagnosing system indexing. It shows how completely and correctly your project's data is indexed in the search engines, and helps you quickly find the cause when search returns only some of the records or lags behind.

To open it, click the index icon in the top bar (header). The icons in the top bar are described in basic interface elements.

Index diagnostics window: "Table Health" with the indexing queue, plus "Data Consistency", "Materialized Views", "Elasticsearch" and "Vector Data Base Coverage" tables

info

A difference between "Expected" and "In search" is normal during reindexing or right after you edit attribute sets: the data catches up at the next scheduled indexing. If the gap remains after the next indexing, there is a problem.

Table Health​

The top block shows the overall indexing state across all system tables.

  • Total: N/N ✓: how many of the tables are in a correct state.
  • Index-data Queue: the state of the data indexing task queue:
    • Active: tasks running right now.
    • Waiting: tasks waiting to run.
    • Failed: tasks that ended with an error.
    • Delayed: postponed tasks.
  • Failed Background Jobs: the number of background jobs that ended with an error. Expand the block to see details.
  • Delayed background jobs: the number of postponed background jobs. You can also expand it to see details.

Below is a table with the state of each entity. Its columns:

  • Table: the table name (products, pages, blocks, admins, discounts, and so on).
  • Expected: how many records should be in the index.
  • Indexed: how many records are actually indexed.
  • Not in index: how many records are missing from the index.
  • Orphans: orphaned records, which are in the index but no longer exist in the source data.
  • Skipped: records skipped during indexing (for example, 3 of 4 entities means records were skipped for some of the entities; you can expand the row).
  • Last Indexed: when the table was last indexed, or Never Indexed if it has never been indexed.
  • Status: the overall status of the row. ✓ means the table is fine.

Data Consistency​

The Data Consistency block compares record counts across the different storage systems for each entity, so you can see exactly where the data diverges.

Columns:

  • Entity: the entity (products, pages, blocks).
  • Catalog: the number of records in the main catalog (the data source).
  • Overview Table (views): the number of records in the overview table (materialized views).
  • Search Engine (Elastic): the number of records in Elasticsearch.
  • Smart Search (Vector DB): the number of records in the smart search vector database. A dash (—) means the entity doesn't take part in smart search.

If the values in a row don't match, the differences are highlighted in color, which shows which storage has incomplete data.

Materialized Views​

The Materialized Views block lists the materialized views (overview tables) and the number of rows in each: products_view, pages_view, blocks_view, and so on.

info

On plans with Elasticsearch, view refresh is disabled, so the data in this block may lag behind the actual state.

Elasticsearch​

The Elasticsearch block lists the search indexes and the number of documents in each (products_v1, pages_v1, blocks_v1, and so on). Use it to check how many documents the search engine actually holds for each entity.

Vector Data Base Coverage​

The Vector Data Base Coverage block shows how much of the data the smart search vector database covers. For example, 1/1 (100%) means all eligible records are indexed in the vector DB. A low coverage percentage means some data hasn't reached smart search yet.

Refresh​

The Refresh button in the upper-right corner of the window updates all diagnostic metrics so you can see the current indexing state without reloading the page.

Last updated on