Support ebpf

[fire]

Is your feature request related to a problem?

Given docker containers that contain a c++ game server executable, I don't want to instrument to get observability in SigNoz.

Here's a detailed problem:

In Godot Engine, we have a VOIP server in a docker container. There are clicks and pops in the opus compression in the voip layer.

We want to get an idea of occurrences of these failures but to solve the quality issues require observability.

For SigNoz to help, the solution requires a open telemetry c++ sdk and random places in the docker container which is a lot of effort to evaluate SigNoz.

If via eBPF, we can runtime insert probes and observe without recompiling, that would make the debugging easier.

To gain observability, it may not be possible to log via instrumentation efficiently enough and would requires a kernel module like ebpf.

Describe the solution you'd like

Support ebpf in SigNoz for observability.

Describe alternatives you've considered

1. https://github.com/cilium/hubble
2. https://github.com/Flowmill/flowmill-collector
3. https://github.com/pixie-labs/pixie
4. https://github.com/cloudflare/ebpf_exporter
5. https://www.percona.com/blog/2020/07/06/adding-ebpf-based-metrics-to-percona-monitoring-and-management/
6. https://docs.projectcalico.org/about/about-ebpf

Additional context

I believe it's possible to make an ebpf collector demo via https://github.com/Flowmill/flowmill-collector and the OpenTelemetry Collector.

[fire]

The authority on ebpf wrote an article on how to add eBPF. https://brendangregg.com/blog/2021-07-03/how-to-add-bpf-observability.html

[nemccarthy]

Looks like there is now a native open telemetry eBPF component that has been forked from Flowmill: https://github.com/open-telemetry/opentelemetry-ebpf

[nacardin]

This would be helpful for monitoring network metrics such as # of active TCP connections in a pod.