I am injecting the spanId and traceId flowing via HTTP header to to build a remote context and creating new Span, however the new span does not get the parent trace and span but created its own trace id. As a result it does not get attached to the desired parent span.

            String[] ids = (headers.get("x-cloud-trace-context").split(";"))[0].split("/");
            logger.info(String.format("Trace Parent Context  '%s' ", ids[0]));
            logger.info(String.format("Span Parent Context  '%s' ", ids[1]));

            SpanContext remoteContext = SpanContext.createFromRemoteParent(
                    ids[0],
                    ids[1],
                    TraceFlags.getSampled(),
                    TraceState.getDefault());

            Span span = tracer
                    .spanBuilder("Make Call Out")
                    .setParent(Context.current().with(Span.wrap(remoteContext)))
                    .startSpan();

                Span childSpan = tracer.spanBuilder("HTTP Request")
                        .setAttribute("method", request.method())
                        .setAttribute("url", request.uri().toString())
                        .setParent(Context.current().with(span))
                        .startSpan();
//Some Work
                childSpan.end();
            span.end();

When I see trace explorer the default trace (from Cloud Run endpoint invocation ) and custom trace (from Java code) has two different traces. Ideally they should be hooked in same hierarchy in single trace.

Related Example