-
|
Context I’m working with a setup where a dispatcher triggers multiple agents. Expected Behavior When the dispatcher sends an AgentExecutorRequest to all agents, the system works as expected: Each agent returns an AgentExecutorResponse. The aggregator receives all responses and produces the final aggregated output. Actual Behavior When the dispatcher triggers only a subset of agents, the aggregator doesn’t work properly — it doesn’t seem to aggregate or produce the final result. Question Is this behavior by design (i.e., the aggregator expects responses from all agents), or is there likely an issue in my implementation or configuration? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Yes. Fan in edges requires all inputs to be available before the target executor is triggered. Instead of using Fan in edges, you can instead wire the edges individually, and handles each input separate and aggregate them in your implementation. That will give you the behavior you want. |
Beta Was this translation helpful? Give feedback.
Yes. Fan in edges requires all inputs to be available before the target executor is triggered.
Instead of using Fan in edges, you can instead wire the edges individually, and handles each input separate and aggregate them in your implementation. That will give you the behavior you want.