feat(querier): add wildcard tenant query support

[amogh-orolabs-dev]

Summary

This PR adds experimental support for wildcard tenant queries in Loki, allowing administrators to query all tenants using * in the X-Scope-OrgID header instead of explicitly listing each tenant ID.

Closes #7356

Motivation

In multi-tenant Loki deployments with many tenants, administrators currently need to manually maintain lists of tenant IDs in their datasource configurations or query headers. This is error-prone and requires constant updates as tenants are added or removed.

This feature addresses the community request for a way to query "all tenants" or "all tenants except specific ones" without explicit enumeration.

Changes

New Files

• pkg/querier/tenant_discovery.go - Core tenant discovery interface and storage-based implementation
• pkg/querier/wildcard_tenant_querier.go - Querier wrapper that handles wildcard resolution
• pkg/querier/tenant_discovery_test.go - Unit tests for tenant discovery
• pkg/querier/wildcard_tenant_querier_test.go - Unit tests for wildcard querier
• integration/wildcard_tenant_queries_test.go - Integration tests
• docs/sources/configure/wildcard-tenant-queries.md - User documentation

Modified Files

• pkg/querier/querier.go - Added configuration options:
  • wildcard_tenant_queries_enabled (bool, default: false)
  • wildcard_tenant_cache_ttl (duration, default: 5m)
• pkg/loki/modules.go - Wired wildcard querier into initialization

Design

Tenant Discovery

The feature uses a TenantDiscovery interface:

type TenantDiscovery interface {
    GetAllTenants(ctx context.Context) ([]string, error)
    IsWildcard(orgID string) bool
}

The StorageTenantDiscovery implementation discovers tenants by listing them from the index storage (S3/GCS/etc). Results are cached with a configurable TTL to avoid expensive storage operations on every query.

Query Flow

1. Request arrives with X-Scope-OrgID: * (or *|!excluded)
2. WildcardTenantQuerier.resolveWildcardTenants() detects the wildcard
3. StorageTenantDiscovery.GetAllTenants() retrieves all tenant IDs from storage (or cache)
4. Exclusions are applied if any !tenant patterns are present
5. Context is updated with resolved tenant IDs
6. Query proceeds through normal MultiTenantQuerier code path
7. Results include __tenant_id__ label on each log entry

Supported Syntax

Syntax	Description
*	Query all tenants
*|!tenant1	All tenants except tenant1
*|!t1|!t2|!t3	All tenants except t1, t2, t3

Configuration

querier:
  multi_tenant_queries_enabled: true
  wildcard_tenant_queries_enabled: true
  wildcard_tenant_cache_ttl: 5m

Testing

• Unit tests for tenant discovery caching, concurrent access, edge cases
• Unit tests for wildcard resolution, exclusion handling, error cases
• Integration tests for end-to-end validation

Breaking Changes

None - this is an opt-in experimental feature.

Checklist

• Tests added
• Documentation updated
• No breaking changes
• Implementation fully wired into Loki startup

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 67afe0b524

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Avoid treating valid tenant IDs starting with ! as exclusions

With wildcard queries enabled, any tenant ID that starts with ! is now parsed as an exclusion and triggers ErrExcludeWithoutWildcard unless * is present. Loki’s tenant validation allows ! in tenant IDs, so a tenant named like !prod would become unqueryable once this feature is turned on. That’s a regression for valid tenant IDs and means enabling wildcard queries can break existing tenants that start with !.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Discover tenants from all schema configs, not just active

Tenant discovery is tied to only the active schema period’s object store and prefix. In deployments with multiple schema configs (e.g., a migration that changed index store or prefix), tenants that only have data in non-active periods won’t be discovered, so * queries will omit them even when the query time range includes their historical data. That violates the “all tenants” expectation for wildcard queries in multi-schema setups.

Useful? React with 👍 / 👎.