Hi team, I’m running Arize Phoenix with the autogen agent framework, and I’ve noticed that each test run shows some unexpected “unknown” spans in addition to the normal chain spans. I’m not sure if this is due to the code structure or configuration. Do you have any recommendations on how to investigate or clean them up? thank you!
Hi Tony, Thanks for the quick response! I updated to Phoenix v12.6.1, but I'm still encountering the same error in the Sessions UI
Error fetching GraphQL query 'sessionLoaderQuery' with variables '{"id":"0"}':
[{"message":"an unexpected error occurred","locations":[{"line":4,"column":3}],"path":["session"]}]current versions:
- arize-phoenix: 12.6.1
- arize-phoenix-client: 1.21.0
- arize-phoenix-otel: 0.13.0
- openinference-instrumentation: 0.1.40
- openinference-instrumentation-autogen-agentchat: 0.1.5
- openinference-instrumentation-openai: 0.1.39
- openinference-semantic-conventions: 0.1.23and my implementation follows the pattern:
tracer = trace.get_tracer(__name__)
with tracer.start_as_current_span(
"query_request",
attributes={
SpanAttributes.OPENINFERENCE_SPAN_KIND: OpenInferenceSpanKindValues.CHAIN.value,
SpanAttributes.SESSION_ID: session_id,
SpanAttributes.INPUT_VALUE: message,
}
) as root_span:
with using_session(session_id):
# Child spans here
root_span.set_status(Status(StatusCode.OK))
```
The traces appear correctly in Phoenix, and I can see the session.id attribute, but the Sessions page shows the GraphQL error. Could you provide a minimal sample script to help me verify if the issue is with my code or still on the Cloud,
😀
Hi team, I'm working on setting up Arize Phoenix monitoring for our autogen-based agent service that uses websocket connections, and running into some issues with the cloud configuration - cannot see my project in the phoenix cloud dashboard.
current configuration
# This works (Local)
# Option 1: Local Phoenix
os.environ["PHOENIX_COLLECTOR_ENDPOINT"] = "http://127.0.0.1:6006"
tracer_provider = register(
project_name="thinkspace-agent-monitoring",
endpoint="http://127.0.0.1:6006/v1/traces",
auto_instrument=True
)
this doesnt work (cloud)
# Option 2: Cloud Phoenix
# if not os.getenv("PHOENIX_API_KEY"):
# raise RuntimeError("PHOENIX_API_KEY not set in environment or .env file")
# tracer_provider = register(
# project_name="thinkspace-agent-monitoring",
# endpoint="https://app.phoenix.arize.com/s/an-qi-yu/v1/traces",
# auto_instrument=True
# )
our websocket notification flow
@app.include_router(ws_chat_notification_router) # WebSocket endpoint
@app.include_router(query_router) # HTTP query endpoint
# E2E test shows local monitoring works great:
with using_session(session_id):
with tracer.start_as_current_span("e2e_test_full_flow"):
# HTTP request triggers agent
# Agent sends WebSocket notifications
# Want to see full trace in Phoenix Cloud