Issues with Using @tracer.chain Decorators in Phoenix App | Arize AI Community

Arize AI Community Icon

RunLLM
APP
·
It seems likely that you’re running into this issue because the tracer you’re fetching via trace.get_tracer(name) is a ProxyTracer (the global default) that hasn’t been “upgraded” with Phoenix’s extra methods such as chain. Instead, after calling register(), try getting your tracer directly from the returned tracer provider. For example, modify your retrieval to: LBKSfrom phoenix.otel import register tracer_provider = register() # Optionally: register(protocol="http/protobuf") tracer = tracer_provider.get_tracer(__name__)LBKSThen use @tracer.chain on your function. This should give you a tracer instance that includes the chain decorator. Does this resolve your issue? Are you able to share a more complete snippet to confirm that all steps are correctly applied?---Sources: - Phoenix Tracing Documentation
💡 Hint: Mention RunLLM in the thread for followups.
Hakan T.
·
For the response above by RunLLM, the decorator and register function are in different files so I can not really do that? RunLLM
RunLLM
APP
·
Make sure you're using the same tracer provider instance across both files. When you register your tracer with Phoenix in one file, export that instance so that other modules can import it. For example: tracer_provider.py LBKSfrom phoenix.otel import registerLBKS# Register and export the Phoenix tracer provider tracer_provider = register()LBKS# Optionally set it as the global provider from opentelemetry import trace trace.set_tracer_provider(tracer_provider)LBKSapi_functions.py LBKSfrom tracer_provider import tracer_providerLBKS# Obtain the tracer from the shared instance—not via trace.get_tracer() tracer = tracer_provider.get_tracer(__name__)LBKS@tracer.chain def my_function(arg): # Function implementation here return argLBKSBy importing the shared tracer_provider from a common module, you ensure that the tracer you get has the extra Phoenix methods (like chain) instead of the default ProxyTracer.Does this approach work for your code structure? Would you like further details on organizing your project? ---Sources: - https://arize-ai.slack.com/archives/C04R3GXC8HK/p1723961775.432439 - https://arize-ai.slack.com/archives/C04R3GXC8HK/p1724796536.922089
💡 Hint: Mention RunLLM in the thread for followups.
Xander S.
·
Hey Hakan T., go ahead and DM me and we'll debug together
👍1