Open Tracing Integration
sebastianharko opened this issue · comments
Here is a potential breakdown of the work that has to be done to make tracing "usable":
- Ensure that what we see in Zipkin (or other ...) mirrors the sequence diagram that Gary created.
- The spans should include sufficient information to be useful (for example, the command/message name should be included (i.e. "AddItem" or "GetState")
- Tracing of actor messaging should include not only execution time (i.e. the code in the "receive" block) but also network time and time that the message in the mailbox. Currently, the abstraction in use (ActorWithTracing) only cares about execution time (i.e. the time it takes to execute the code in the 'receive' block)
- Tracing should work when Surge is deployed as 'cluster' on Kubernetes
Closing this since now both the consumer and the command side are instrumented