cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

can we change the secuence of calls of a trace from distributed traces ?

MariettaAlmeida
Organizer

hi there

when we are using distrubutes traces from a particular service, we've noticed that the flow of calls made from a trace is not complete. 

for example, we have the service A who calls the service B and then call an unmonitored host. when using the distributed traces section we see request from the service A calling B or service A calling the unmonitored host but we dont have the complete trace. is there a way to modify the topology of the trace so we can have the complete call instead of parts of it ?

we want to see on the distributes traces thas "califimmm" calls "nlbcbbbb" and then to the opaque host instead of "califimmm" either calling nlbcbbbb" or the opaque host

 

2 REPLIES 2

JoseRomero
Dynatrace Advisor
Dynatrace Advisor

Hi @MariettaAlmeida ,

To see the complete trace, you need to instrument the service on the unmonitored hosts. You have several options:

  • Install OneAgent: The unmonitored host should have OneAgent installed and configured in full-stack and deep monitoring to create the services and allow you to see the complete trace.
  • Use OpenTelemetry: Instrument the services on the unmonitored host with OpenTelemetry and send the traces to Dynatrace.

The most recommended approach is to identify those unmonitored hosts and install OneAgent on them for monitoring. If the unmonitored host is an external service, you can create a custom service to provide context to that service.

In summary, you need to install OneAgent in full-stack mode and ensure deep monitoring to achieve full observability and complete your trace.

Julius_Loman
DynaMight Legend
DynaMight Legend

Besides what @JoseRomero mentions, there can be other reasons. Tracing works on trace context propagation. There can also be unsupported technologies where just deploying OneAgent does not help or there can be also applications which are written in a way where trace context cannot be propagated automatically. An example is a custom messaging implementation or thread-to-thread communication within the same process (producer/consumer thread pools). Those situations require minor code changes and the use of OpenTelemetry or OneAgent SDK in the code to fully propagate the context.

Certified Dynatrace Master | Alanata a.s., Slovakia, Dynatrace Master Partner

Featured Posts