Skip to content

Latest commit

 

History

History
52 lines (42 loc) · 2.84 KB

README.md

File metadata and controls

52 lines (42 loc) · 2.84 KB
Raw

Configuration

Broker and topics

Configuration Description Default
KAFKA_BOOTSTRAP_SERVERS Kafka bootstrap servers localhost:9092
KAFKA_SPANS_TOPIC Topic where incoming list of spans are stored. zipkin-spans
KAFKA_TRACE_TOPIC Topic where aggregated traces are stored. zipkin-trace
KAFKA_DEPENDENCY_TOPIC Topic where aggregated service dependencies names are stored. zipkin-dependency

These topics can be configured by individual components (span-consumer, aggregation, storage) by using Java options. For more information about property names, check zipkin-server-storage-kafka.yml

Storage configurations

Configuration Description Default
KAFKA_STORAGE_HOSTNAME Host name used by storage instances to scatter-gather results localhost
KAFKA_STORAGE_DIR Root path where Zipkin stores tracing data /tmp/zipkin-storage-kafka
KAFKA_STORAGE_PARTITIONING_ENABLED Flag to enable Span partitioning. true
KAFKA_STORAGE_AGGREGATION_ENABLED Flag to enable Span aggregation. true
KAFKA_STORAGE_AGGREGATION_TRACE_TIMEOUT How long to wait until a trace window is closed (ms). If this config is to small, dependency links won't be caught and metrics may drift. 600000 (1 minute)
KAFKA_STORAGE_TRACE_ENABLED Flag to enable Trace storage. true
KAFKA_STORAGE_TRACE_TTL How long to keep traces stored. 259200000 (3 days)
KAFKA_STORAGE_TRACE_TTL_CHECK_INTERVAL How often check traces stored TTL. 3600000 (1 hour)
KAFKA_STORAGE_DEPENDENCY_ENABLED Flag to enable Dependency storage. true
KAFKA_STORAGE_DEPENDENCY_TTL How long to keep dependencies stored. 604800000 (1 week)

When running on Kubernetes/Openshift

When running on Kubernetes/Openshift is recommended to use statefulsets in order to maintain storage state directories.

Configure hostname

For instances to access other pods on the stateful set, we have to use valid DNS-names:

          env:
            - name: STORAGE_TYPE
              value: kafka
            # Gather hostname (name-${POD_ID}) from metadata
            - name: HOSTNAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            # Mapping hostname to Kubernetes DNS defined service name (${NAME}-${POD_ID}.${SVC}.${NAMESPACE}.svc.cluster.local),
            # then instance storage becomes accessible between them
            - name: KAFKA_STORAGE_HOSTNAME
              value: $(HOSTNAME).zipkin.default.svc.cluster.local