Logs not nested within HTTP request main log

[edouardmercier]

This issue is related to issues #674 and #677 and to the documentation https://docs.quarkiverse.io/quarkus-google-cloud-services/main/logging.html#quarkus-google-cloud-logging_quarkus-google-cloud-logging-gcp-tracing-project-id.

When using version v2.17.1 of the quarkus-google-cloud-logging component, it seams that no default implementation of the io.quarkiverse.googlecloudservices.logging.runtime.TraceInfoExtractor and that you need to set up one in order to provide the trace and the spanId. I'm resorting to the org.jboss.logmanager.LogManager for logging.

I have implemented the TraceInfoExtractor in the hope to extract from the provided ExtLogRecord object the trace and the spanId of the HTTP request resulting from the web service endpoint invocation, but its class does not expose that information (the only field which may make sense if sequenceNumber). Hence, when logging during the execution of a web service, the output log is not nested in Google Cloud Logging within the HTTP log request.

Did I miss something? Is it possible to extract from the context the HTTP request trace and the spanId so that I can implement the TraceInfoExtractor` properly by reusing them for the logs issued during its execution? Do you have plans to provide such default implementation (it would save hours to newcomers like me)?

[jfbenckhuijsen]

Hi @edouardmercier, I've been working a bit on this and currently use the following setup:

(the code is in kotlin, but I think you should get the point, once we're clear on whats needed ill create a PR to incorporate this into the main library)

@ApplicationScoped
class LoggingTraceProvider: TraceInfoExtractor {

    override fun extract(record: ExtLogRecord?): TraceInfo? {
        val spanContext = Span.current().spanContext
        return if (spanContext.isValid) TraceInfo(
            spanContext.traceId,
            spanContext.spanId
        ) else null
    }
}

this is combined with the following logging setup:

quarkus:
  google:
    cloud:
      logging:
        enabled: true
        format: JSON
        default-log: my-default-log
        resource:
          type: cloud_run_revision
          label:
            configuration_name: ${K_CONFIGURATION:my-service}
            location: ${GCP_REGION:local}
            revision_name: ${K_REVISION:my-service-local}
            service_name: ${K_SERVICE:my-service}
        structured:
          stack-trace:
            included: true
        gcp-tracing:
          enabled: true
          project-id: ${GCP_PROJECT}
        log-target: CLOUD_LOGGING

This setup is for a cloud run project of course, but should be adaptable to whatever you aare running. Lemme know if this fulfills the nesting of the logs in the main log, then Ill get a PR running to incorporate this