QUERY_TIMEOUT: Query Request Was Running Too Long

What does this error mean?

Salesforce enforces a server-side query execution timeout to protect the multi-tenant platform. When a query takes too long — typically because it requires a full table scan on a large object — Salesforce terminates it and returns this error. The timeout threshold is not publicly documented but is generally triggered by non-selective queries on objects with hundreds of thousands or millions of records.

This error is different from the Apex CPU time limit — QUERY_TIMEOUT is a database-level timeout measured in wall-clock time, while CPU time is the time Apex spends in computation.

Common Causes

1. Full table scan on a large object

A query with no WHERE clause, or a WHERE clause that filters only on non-indexed fields, forces Salesforce to read every row in the table.

2. Complex WHERE clause on unindexed fields

Multiple non-indexed conditions combined with OR logic, or negation operators (!=, NOT IN), that cannot use any index and require scanning the entire table.

3. Reporting on objects with millions of records

Ad-hoc queries for reporting purposes run against production-sized data without the selective filters that operational queries use.

How to Fix It

Solution 1: Add indexed field filters

-- ❌ Full table scan
SELECT Id, Name FROM Case WHERE Custom_Category__c = 'Support'

-- ✅ Indexed fields narrow the scan
SELECT Id, Name FROM Case
WHERE CreatedDate = LAST_N_DAYS:90   -- indexed
  AND OwnerId = :currentUserId        -- indexed
  AND Custom_Category__c = 'Support'  -- now filtered on a smaller set

Solution 2: Request a custom index

If your application repeatedly filters on the same non-indexed custom field, raise a Salesforce Support case requesting a custom index. Salesforce evaluates the field's selectivity and adds the index if approved — free of charge.

Solution 3: Use Batch Apex for large-data reporting

For analytics queries that inherently need to scan large datasets, restructure as Batch Apex with Database.getQueryLocator() and process results in manageable chunks.