feat(sdk-logs): emit resource attributes from ConsoleLogRecordExporter

[harelmo-lumigo]

Which problem is this PR solving?

Logging via the SDK emits the resource attributes, but ConsoleLogRecordExporter does not reflect that - which is very confusing and makes it hard to troubleshoot logging / resource-detection issues.
This PR adds the log-record's resource to the output that's logged to the console to aid troubleshooting.

Fixes #4645

Short description of the changes

Adding a new resource attribute to each log record emitted by ConsoleLogRecordExporter

Type of change

• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

By extending the current tests for ConsoleLogRecordExporter

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: harelmo-lumigo / name: Harel Moshe (7868ae5, a6762a0, 8988a83, 8a6db52, 2a6bafa, c835c29)
• ✅ login: pichlermarc / name: Marc Pichler (67ba523)

[hectorhdzg]

Well my main concern is the same mentioned for Spans by @Flarna in #4605 (comment), this could be particularly bad and noisy in Logs, as this kind of telemetry could be way bigger in volume than spans, having this configurable would be ideal in my opinion.

[harelmo-lumigo]

@hectorhdzg thanks for the feedback.

Just to make sure I got it, by:

having this configurable would be ideal in my opinion.

you mean introducing a new env-var to control the inclusion of resource attributes in each log? or a config-option in ConsoleLogRecordExporter's constructor?

[pichlermarc]

@harelmo-lumigo I'd be inclined to use the same approach as for the trace exporter (keeping your PR as-is) here and see if the increased verbosity poses an actual problem for users. We should not go for differing console exporter interfaces unless necessary (like it is with the metrics exporter where we need temporality configuration for debugging purposes).

If we see that it's a problem and users are irritated by it, I'll open a PR to introduce the option that @trentm suggested in #4645 - we can then continue the discussion on that PR.

[pichlermarc]

@harelmo-lumigo thanks for working on this. To proceed, please follow the instructions on #4646 (comment) to sign the CLA 🙂

[hectorhdzg]

@pichlermarc the way I see ConsoleExporters to be used are for local debugging for developers, when usually you don't have a lot of data flowing so this make perfect sense, but this is also being used as a self troubleshooting mechanism for OPs, where they can enable console exporters through environment variables and look at the actual real data of their services or applications, adding the resource information, that can be huge by the way compared to the actual log record, would make really hard for latest kind of usage. If we don't believe console exporters could be used in that kind of scenario then we should be good though.

[trentm]

If we see that it's a problem and users are irritated by it, I'll open a PR to introduce the option that @trentm suggested in #4645 - we can then continue the discussion on that PR.

Rather this comment: #4605 (comment)
The "ibid" thing might be similarly confusing, however. Not sure.

[hectorhdzg]

Ok I can see there was a long conversation in other PR mentioning exactly my concern, I was just worried this was not considered, I'm fine moving forward with this for now.

[pichlermarc]

Thanks 🙂

Sorry for linking the wrong conversation earlier - as said before I'll continue to monitor user-concerns about having the resource there.

The situations I run into with users (4 times over since September, from actual support-tickets or directly reached out to me via slack while working on a POC) were:

• users think the resource is not attached as they don't see it in the console exporter -> it turns out that everything is actually okay
• users miss a specific resource attribute from detectors and they have no way to troubleshoot it by using a console exporter.

If you get any user feedback regarding this being too verbose please feel free to forward it to me - I'm happy to consider other alternatives but I wanted to go for the most obvious solution first. 🙂