Jaeger Collector Span Ingestion Timeout Behavior During Storage Backend Schema Changes

Span Ingestion Timeout Configuration

The Jaeger Collector handles span ingestion from agents and services, with configurable timeout settings that determine how long the collector waits for storage operations to complete. When the underlying storage backend undergoes schema changes—such as Elasticsearch index template updates or Cassandra schema migrations—the ingestion pipeline may encounter unexpected serialization or query failures.

Timeout Configuration

The collector's collector.timeout flag controls the maximum duration for storage operations. However, this timeout does not account for transient schema incompatibilities that may cause immediate serialization errors rather than delayed responses. The collector returns HTTP 500 errors when spans cannot be processed, but the exact error message distinguishes between timeout conditions and schema-related serialization failures.

Storage Schema Compatibility

Jaeger's documentation indicates that storage schema changes require coordinated application-level migration. The collector does not implement automatic fallback or retry mechanisms for schema-version mismatches. During rolling upgrades of storage backends, spans may be rejected if the collector version expects a schema format that the temporary storage state cannot satisfy.

What is the precise relationship between the Jaeger Collector's ingestion timeout configuration and schema incompatibility errors when spans cannot be serialized to the current storage format? Should the timeout value be adjusted to accommodate schema validation delays, or is there a separate configuration for handling serialization failures during storage backend transitions?