After investigating performance degradation wrt eBPF Agent CPU usage, we found very aggressive GC calls happening in stressed environment (sampling=1 + load injection)

We want to compare performances from two potential fixes:

1. Turning off forced GC by default (https://github.com/netobserv/netobserv-ebpf-agent/pull/184/ )
2. Removing forced GC specifically on the ringbuf tracer ( https://github.com/netobserv/netobserv-ebpf-agent/pull/185 )

Note that 1. can also be achieved by setting in FlowCollector CR under the ebpf section, using the current 1.4 candidate build:

      debug:
        env:
          ENABLE_GARBAGE_COLLECTION: "false"