tracing

Documentation

Index

• func ParseSpanContext(headers map[string][]byte) (sc trace.SpanContext, ok bool)
• func SerializeTrace(spanContext trace.SpanContext) []sarama.RecordHeader
• func StartTraceFromMessage(logger *zap.SugaredLogger, inCtx context.Context, ...) (context.Context, *trace.Span)

Functions

func ParseSpanContext

func ParseSpanContext(headers map[string][]byte) (sc trace.SpanContext, ok bool)

ParseSpanContext takes the "traceparent" and "tracestate" headers and regenerates the trace span context from them. This context can then be used to start a new span that uses the same trace ID as a different (but related) span.

func SerializeTrace

func SerializeTrace(spanContext trace.SpanContext) []sarama.RecordHeader

SerializeTrace returns the traceparent and tracestate values from a span context as a slice of sarama.RecordHeader structs that can be appended to an existing sarama.ProducerMessage

func StartTraceFromMessage

func StartTraceFromMessage(logger *zap.SugaredLogger, inCtx context.Context, message *protocolkafka.Message, topic string) (context.Context, *trace.Span)

StartTraceFromMessage extracts the headers from a message (traceparent and tracestate) and uses them to start a span, which can be used with whatever tracing backend is set up (e.g. Zipkin) in order to trace the flow of a message. Multiple spans may be part of a single trace, for example, a dead letter message or a reply should be easy to match with the original message based on the trace ID. This ID is originally set in the first message header using the SerializeTrace function.