How to Instrument a Langgraph Workflow for Clear Tracing in Phoenix Dashboard

I autoinstrumented my langgraph workflow using phoenix's openinference. After running the workflow a couple of times, I have opened my project in phoenix's web console. I was seeing all traces jumbled up and it was very hard to understand what was going on. I got to know that a trace contains multiple spans. How do I instrument my workflow such that each execution of workflow creates a single trace on the phoenix dashboard and when I click that trace, I will be able to see all the executions and LLM calls done in nodes in the form of spans?