Closed Shackelford-Arden closed 2 years ago
This looks more like an issue with grpc under specific conditions. Can you share some simple reproducible example?
Apologies on the delay here, took me a little longer to put something together that matched how we were running things. Lovingly, when I tried to replicate it outside of our application, I couldn't. Then for sanity I tried replicating it again in our application and I couldn't. So I'm just crazy apparently.
If I see it again and can replicate, I'll come back. For now, I'll close this out.
Describe your environment
Python Version: 3.8.12 ('Linux-3.10.0-1160.62.1.el7.x86_64-x86_64-with-glibc2.2.5')
OpenTelemetry depdenencies:
This project is using Celery and the message that we're finding is only experienced in the Celery worker.
We have the following environment variables set within the container (Docker):
Steps to reproduce
Describe exactly how to reproduce the error. Include a code sample if applicable.
We've been able to reproduce the error every time we initiate work within a Celery task. The project is a bit large, so I'll come back with a smaller scale example.
By the time we get this first log message, we've generated about 50 spans.
What is the expected behavior?
I expected spans to be exported without receiving a log message from the exporter showing up.
What is the actual behavior?
We receive the following exception at what appear to be random times:
I don't know if this is helpful to distracting, but in playing around I did set the GRPC_VERBOSITY to debug. Only on the first occurrence of the above exception did GRPC log anything at the same time:
Note: This GRPC logging does not occur again
Additional context
These log messages are only observed when upgrading the
opentelemetry-sdk
package to 1.7.1.1.6.2 works without log messages appearing.
This particular project is not able to go above Python 3.8.x, but it is noted that another project that has the exact same architecture but is running on Python 3.9.x doesn't have these log messages, even if we go to the latest version of the API/SDK packages.
Based on the context of when we're seeing these log messages, my guess is it might be related to some of the work that was completed in #2242 .