Usage → Failures
Monitor and analyze failed BigQuery queries to reduce wasted resources and improve reliability.
The Failures tab provides visibility into failed queries in your BigQuery environment. Understanding query failures helps reduce wasted resources and improve pipeline reliability.
Overview
The Failures view enables you to:
- Monitor the volume and cost of failed queries
- Distinguish between deterministic and non-deterministic failures
- Identify users and queries with high failure rates
- Drill down into specific failure reasons
Page Controls
Project Filter
Filter metrics to specific BigQuery projects.
Cost Toggle
Toggle between resource metrics (slots) and cost estimates (USD).
Failure Type Filters
Filter by failure type:
- Deterministic: Failures that will always occur (syntax errors, permission issues)
- Non-Deterministic: Failures that may succeed on retry (internal errors, timeouts)
Failure Types
Deterministic Failures
Issues that require code or configuration changes to resolve:
| Error Type | Description |
|---|---|
| notFound | Table or resource not found |
| invalidQuery | Invalid SQL syntax |
| accessDenied | Permission denied |
| invalid | Invalid request or schema mismatch |
Non-Deterministic Failures
Transient issues that may succeed on retry:
| Error Type | Description |
|---|---|
| internalError | BigQuery internal error |
| backendError | Temporary service unavailability |
| rateLimitExceeded | Too many concurrent requests |
Summary Cards
Failed Query Count
Shows the count of failed queries as a percentage of total queries for the last 24 hours, 7 days, and 30 days.
Execution Times
Shows execution time consumed by failed queries as a percentage of total.
Total Slots / Cost
Shows slot consumption or cost from failed queries as a percentage of total.
Failures by Dimension
View failures broken down by:
- By User: Which users have the most failed queries
- By Error Type: Failures grouped by error category
- By Table: Which tables are associated with failures
- By Query Pattern: Which query patterns fail most often
Use Cases
Reducing Wasted Resources
- Review the cost of failures
- Identify top users or queries with high failure costs
- Prioritize fixing failures with the highest resource consumption
Improving Pipeline Reliability
- Focus on deterministic failures to find code/config issues
- Review failures by error type to identify systematic problems
- Work with pipeline owners to fix root causes
Permission Management
- Filter to access denied errors
- Identify users with permission issues
- Update IAM policies as needed
Common Failure Patterns
"Table not found"
- Table was deleted or renamed
- Typo in table reference
- Dataset permissions changed
"Access Denied"
- Missing table-level permissions
- Service account not authorized
- Row-level security blocking access
"Internal Error"
- BigQuery service issues
- Resource contention
- Consider implementing retry logic
Updated 16 days ago