Connection Timeout Error with telemetry.crewai.com

[leeeex]

Hello , I hope this message finds you well. I am reaching out to report an issue that I encountered while using your project. Description: I encountered a ConnectTimeout error when attempting to connect to telemetry.crewai.com on port 4318. The error message is as follows:

requests.exceptions.ConnectTimeout: HTTPConnectionPool(host='telemetry.crewai.com', port=4318): Max retries exceeded with url: /v1/traces (Caused by ConnectTimeoutError(<urllib3.connection.HTTPConnection object at 0x7f34caf38280>, 'Connection to telemetry.crewai.com timed out. (connect timeout=10)')

[Mirgiacomo]

Follow, same error:

raise MaxRetryError(_pool, url, reason) from reason # type: ignore[arg-type].

WARN urllib3.exceptions.MaxRetryError: HTTPConnectionPool(host='telemetry.crewai.com', port=4318): Max retries exceeded with url: /v1/traces (Caused by ConnectTimeoutError(<urllib3.connection.HTTPConnection object at 0x1378487c0>, 'Connection to telemetry.crewai.com timed out. (connect timeout=10)')).

[ansunsujoe]

Related, but I am not able to run CrewAI due to the fact that it hits telemetry.crewai.com which I can't access through my company's proxy. It would be nice if the user could choose to optionally disable telemetry.

[lm150]

Please fix this. I'm in the same situation as telemetry.crewai.com is blocked for my company as well.

[slaughters85j]

I cant even hit it in my local environment with telnet.

% telnet telemetry.crewai.com 4318
Trying 10.0.0.1...
telnet: connect to address 10.0.0.1: Connection refused
Trying fd6e:6c73:b178:1::1...
telnet: connect to address fd6e:6c73:b178:1::1: Connection refused
telnet: Unable to connect to remote host

I also get this about a hundred times during runtime:

requests.exceptions.ConnectTimeout: HTTPConnectionPool(host='telemetry.crewai.com', port=4318): Max retries exceeded with url: /v1/traces (Caused by ConnectTimeoutError(<urllib3.connection.HTTPConnection object at 0x7f34caf38280>, 'Connection to telemetry.crewai.com timed out. (connect timeout=10)')

[chrissonntag]

I agree to the previous speaker. Their must be a setting to turn those requests off. You can leave them on by default but allow disabling them for good. In best case it'd be a env var so that we do not need to do it over and over again.

[eleqtrizit]

I want to add to this.

[heyfixit]

it's not pretty, but a tiny bit of monkey patching might get ya there. Mind the typo on crewai.telemtry at the moment (have a PR to fix that one: https://github.com/joaomdmoura/crewAI/pull/295 ).

from crewai.telemtry import Telemetry

def noop(*args, **kwargs):
    # with open("./logfile.txt", "a") as f:
    #     f.write("Telemetry method called and noop'd\n")
    pass


for attr in dir(Telemetry):
    if callable(getattr(Telemetry, attr)) and not attr.startswith("__"):
        setattr(Telemetry, attr, noop)

[psikosen]

Bump, still is happening on my side, has a fix been pushed to pip ?

[vipulnath]

I am also facing the same issue even after this.

[chrissonntag]

Issue #241 has been closed without any comments. We can assume at this point that there is no interest what so ever to implement a flag to turn of the telemetry calls.

[psikosen]

Yup np, I notice the issue doesn't happen while using Ollama, only while using lmstudio. So it's fine to close.

[eleqtrizit]

It's not fine to close. It happens when the telemetry url is being blocked by a firewall. What you're using has nothing to do with the whether it is OK or not.

It does look like there is a new flag, 'telemetry' for the crew to disable.

[pudimKBM]

Hey i have a fix for this you can just disable the telemetry calls on the tool usage class and in the crew class

Like below C:/User/.../your-env/site-packages/crewai/crew.py C:/User/.../your-env/site-packages/crewai/tools/tool_usage.py

You can comment all the telemetry related stuff and should work fine again

[mabry1985]

bumped, this is an issue still and needs to be resolved. Add an opt out completely option.

[gardner]

I just published a pypi package from my opt-out branch: https://pypi.org/project/crewai-clean/

You need to set CREWAI_TELEMETRY_OPT_OUT=True in your environment.

Please test it out and leave a comment on https://github.com/joaomdmoura/crewAI/pull/402

[alexjst]

I just hit the same problem. I disabled telemetry with one line change to the file "<python_lib_dir>/site-packages/crewai/telemetry/telemetry.py" shown as diff below:

- self.ready = True
+ self.ready = False

Indeed, crewai should probably provide an option to allow clients to opt out of telemetry. In my case, the company's security policy wouldn't allow this without legal process and approval.

[Yamijalaa]

I faced the same issue as above. Please help me in resolving this:

requests.exceptions.ConnectTimeout: HTTPConnectionPool(host='telemetry.crewai.com', port=4318): Max retries exceeded with url: /v1/traces (Caused by ConnectTimeoutError(<urllib3.connection.HTTPConnection object at 0x000002311B65F310>, 'Connection to telemetry.crewai.com timed out. (connect timeout=10)'))

[gardner]

@Yamijalaa

Set the env var OTEL_SDK_DISABLED=true

import os
from crewai import Agent, Task, Crew, Process
from crewai_tools import SerperDevTool

os.environ["OTEL_SDK_DISABLED"] = "true"

[johnisanerd]

@gardner , I did this as well, just slightly differently:

OTEL_SDK_DISABLED = "true"

I get the following message over and over, but the failures were making the output completely unreadable.

2024-05-22 21:49:28,225 - 8099684480 - __init__.py-__init__:1198 - WARNING: SDK is disabled.

[gardner]

@johnisanerd that warning looks like it come from the Open Telemetry Python SDK here.

If you add this to your /etc/hosts file:

127.0.0.1	telemetry.crewai.com

You will not see connection errors.

To test if OTEL_SDK_DISABLED has an effect:

If you then run without OTEL_SDK_DISABLED being set, you will see the connection errors. This indicates the underlying OpenTelemetry SDK is succesfully disabled when setting OTEL_SDK_DISABLED = "true"

[AJNOURI]

I solved the issue by faking the response from the telemetry url:

.../lib/python3.11/site-packages/opentelemetry/exporter/otlp/proto/http/trace_exporter/init.py

from requests.models import Response
    ...

    def _export(self, serialized_data: bytes):
        data = serialized_data
        if self._compression == Compression.Gzip:
            gzip_data = BytesIO()
            with gzip.GzipFile(fileobj=gzip_data, mode="w") as gzip_stream:
                gzip_stream.write(serialized_data)
            data = gzip_data.getvalue()
        elif self._compression == Compression.Deflate:
            data = zlib.compress(serialized_data)

        # return self._session.post(
        #     url=self._endpoint,
        #     data=data,
        #     verify=self._certificate_file,
        #     timeout=self._timeout,
        # )
        # Create a fake response object
        fake_response = Response()
        fake_response.status_code = 200  # Set status code to 200 OK
        fake_response._content = b'{"status": "ok"}'  # Optional: fake response content

        return fake_response  

For now it works like a charm

[beeapu]

I am still facing this issue but with my personal project. r = adapter.send(request, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ ... \crewai_env\Lib\site-packages\requests\adapters.py", line 688, in send raise ConnectTimeout(e, request=request) requests.exceptions.ConnectTimeout: HTTPSConnectionPool(host='telemetry.crewai.com', port=4319): Max retries exceeded with url: /v1/traces (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x000001C2A5AD4170>, 'Connection to telemetry.crewai.com timed out. (connect timeout=30)'))

but when i try to clone this repo: https://github.com/rajib76/multi_agent/blob/main/08_how_to_do_memory_in_crew_01.py and used it but never faced any timeout. They are using same conda environment, and not faced the timeout no matter how many tokens I use

[slimshreydy]

Telemetry went down recently because the crewai.com server itself was experiencing issues, and made it incredibly difficult to develop: https://github.com/crewAIInc/crewAI/issues/1164