@chombium Thanks for the feedback. That sounds like a slightly different feature request. In this PR we're really just trying to ensure that we can reserve the ability to add exporters without risk of conflicting with other exporters defined by the operator.

What you're suggesting sounds interesting to consider for the future though.

@acrmp Thanks for the clarification. I totally agree with what you have written.

I always hated that the application container metrics and application logs are mixed together with the platform component metrics and in the Firehose Architecture, some noisy application can cause drops of other log envelopes including platform metrics. The aggregate Syslog drains fix the noisy neighbor problem, but the consumers still get a mix of all metrics and have to do the filtering themselves.

Therefore with the Otel collector, I'm thinking of separating platform component metrics in their own pipeline, possibly adding some custom processors to do some metrics' content adjustments and sending only platform component metrics to the Platform Monitoring System whatever that is.

Or another thing is on-the-fly inspection of the metrics with a custom processor and sending alerts (via Alerting System) immediately as close to the source of the metrics.

OTel will open another level of possibilities for monitoring CF. I'm really excited about the improvements which we can bring to the logging and metrics stack and the observability of CF. :)