Failure to export from collector to backend (GO Lambda Layer)

[itaybaror]

I have been attempting to instrument lambdas using the docs on the ADOT website. I have successfully instrumented some python lambdas, where they send the data to the collector, and then exports the data to lightstep (my backend). However when i try to instrument in go, i do see the logs in cloudwatch (i configured the collector to send data to lightstep and to logs) but when it tries to export the data to lightstep (and i have tried other backends) i am usually greeted with some sort of grpc or timeout error, and i do not see the data in the backend.

I have attached pictures with the errors i receive. Keep in mind that with my working python implementation i am using the same collector config file.

Errors:

My collector config file:

[fatsheep9146]

I think the error log is pretty clear.

dialing dial tcp 35.222.202.121:443: i/o timeout 

Can you check the network configuration of your collector? For example login to the collector's container or machine, and run

ping 35.222.202.121

or

telnet 35.222.202.121 443

[itaybaror]

That is not always the error, not it is this:

[bogdandrutu]

@carlosalberto @jmacd @tedsuo can you look into this?

@itaybaror can you confirm that this was not a transient error on the backend?

[jangaraj]

For the record: the same issue in another project - https://github.com/aws-observability/aws-otel-lambda/issues/271#issuecomment-1195804948

[stephenhong]

Are there any updates to this issue? I'm experiencing something similar where my lambda functions work fine with otel lambda layers (Java, Node.js, Python) but fails to export traces with the Go lambda layer

[bogdandrutu]

@MrAlias any idea?

[MrAlias]

Hmm, I'm not 100% sure what the issue is here.

The authentication issue seems relevant. I'm guessing either the network connection is being setup without TLS but the server expects this, or the Lightstep authentication token is somehow blocking the connection.

@MadVikingGod thoughts?

[itaybaror]

@bogdandrutu definitely not an issue with the backend, tried with multiple. Also same configuration file works in python lambdas.

[bogdandrutu]

@MadVikingGod @codeboten @paivagustavo any idea?

[codeboten]

@itaybaror can you reproduce this issue using the collector outside of the lambda layer? This could at least remove some of the moving pieces here.

[mx-psi]

Closing as stale since this is 'waiting for author' and we have not heard back since January. Feel free to comment to reopen :)