#support

Issue with Signoz Tracer on Python FastAPI Application

TLDR Snehanshu is experiencing an issue with the Signoz tracer on a Python FastAPI application which involves a large number of irrelevant spans. vishal-signoz suggested verifying the relevance of these spans and questioned about the type of instrumentation used by Snehanshu.

1mo

Join the chat

Aug 17, 2023 (1 month ago)

Photo of md5-a9a351e11d64f05b41fec183816a0cda

Snehanshu

04:24 AM

Hi team, when I am using Signoz tracer to monitor a method in my Python FastAPI application which involves jpype library, there is huge numbr of traces coming up on the UI and also the UI hangs on opening up that particular request. Could anyone help me know why this is happening?

Photo of md5-f936d3e5743d23344d6c60813189716f

vishal-signoz

07:10 AM

How many spans are there?
The community edition supports upto 10k spans in a single trace. The enterprise plan has a feature which allows upto 100k spans in a single trace.

Snehanshu

07:31 AM

vishal-signoz 2998 spans coming up. However i have attached traces only to few methods. These many spans showing up when jpype is used.
Other cases only the required number of spans show up.
Is it normal behaviour or something is wrong from my code end?

vishal-signoz

07:32 AM

Can you check if the spans are relevant?

Snehanshu

07:33 AM

No they are not relevant. they have time spans of 0.01 microseconds.

vishal-signoz

07:33 AM

Our enterprise customers do have thousands of spans so depends on your system.

Snehanshu

07:33 AM

Outside of the main span which is actually required.

vishal-signoz

07:36 AM

Are they duplicates? Can you their name and attributes?

Snehanshu

07:36 AM

If you see this screenshot, the upload span is the only required span, however the POST/ ones are in huge number coming up.

07:36

Snehanshu

07:36 AM

Yes they are duplicates the POST/ ones

vishal-signoz

07:38 AM

I didn’t find any relevant github issues on jpype. How did you instrument?

Snehanshu

07:39 AM

from opentelemetry import trace
tracer = trace.gettracer(_name)
with tracer.start_as_current_span("file_response"):

vishal-signoz

07:39 AM

So you are manually instrumenting?

Snehanshu

07:39 AM

Yes within each method which i want to be monitored. Usig this.
with tracer.start_as_current_span("file_response"):

07:41

Snehanshu

07:41 AM

There is a nested method calls which comes up perfectly as in the screenshot marked with 7, upload.

07:41

Snehanshu

07:41 AM

However i am not able to get why those extra POST spans are showing up.

08:02

Snehanshu

08:02 AM

Also FYI, the API is being run by the command:
OTEL_RESOURCE_ATTRIBUTES=service.name=autofill_exports OTEL_EXPORTER_OTLP_ENDPOINT="endpoint" opentelemetry-instrument <followed by the api command>

vishal-signoz

09:04 AM

Any reason why you are not using auto instrumentation?

Snehanshu

09:44 AM

When using auto instrumentation i was not getting spans for individual methods.

09:44

Snehanshu

09:44 AM

Or could you point me to any correct resource?

09:45

Snehanshu

09:45 AM

using auto instrumentation?

09:53

Snehanshu

09:53 AM

https://signoz.io/docs/instrumentation/python/
on following this page i was getting only the endpoint span, and this was too in large number. This did not give the spans for the methods it was calling internally.

Aug 20, 2023 (1 month ago)

Snehanshu

07:32 AM

vishal-signoz if you could answer this?

SigNoz Community

Built with ClickHouse as datastore, SigNoz is an open-source APM to help you find issues in your deployed applications & solve them quickly | Knowledge Base powered by Struct.AI

Indexed 825 threads (61% resolved)

Join Our Community Manage subscription

Similar Threads

Issues with Signoz 10.2: Grouping, Missing Traces, and Capturing Requests

Anil reported issues with Signoz 10.2, including grouping, missing traces, and capturing requests. Srikanth suggested upgrading to a newer version. The issue was resolved after upgrading the node version to 18 and additional clarifications were provided.

4mo

Filling in Method for Traces in Go Instrumentation

Timothy asked about filling in the method for traces in Go. vishal-signoz identified a workaround by changing the spankind to Internal, and shared a PR for the fix.

7mo

Unable to Monitor jdbc_connections_active Metrics in SigNoz

Ragulraj is unable to monitor jdbc_connections_active metrics with SigNoz despite modifying configurations and adding dependencies. Srikanth provided troubleshooting steps and info on alternatives, but the issue remains unresolved.

5mo

Troubleshooting Django App Instrumentation with Signoz

tim encountered issues instrumenting a Django app with Signoz. tim resolved the problem by identifying the need for a client to forward content to Signoz.

4mo

Troubleshooting SigNoz Auto-Instrumentation Configuration

igor is having trouble configuring auto-instrumentation for Java applications using SigNoz, with traces not appearing in the SigNoz UI. Prashant advises to check logs of the otel sidecar, use service name for endpoint, verify supported libraries, and test with telemetrygen. However, the issue still persists.

5mo