Stale ebpf maps if agent stops abruptly

[anubhabMajumdar]

Describe the bug

If agent pod is OOMKIlled, Packetparser leaves behind stale maps and qdiscs. These are never cleaned up on restart.

To Reproduce
Steps to reproduce the behavior:

1. Deploy retina-advanced
2. Exec into a node and kill the controller process repeatedly
3. Check maps and qdiscs

Expected behavior
Only one instance of maps should exist for each plugin and one ingress/egress qdisc for each veth.

Platform (please complete the following information):

• OS: Linux
• Kubernetes Version: 1.29
• Host: AKS
• Retina Version: current

Additional context
Suggestion - Cleanup should happen in init container (probably we need privilege to clean up residual maps and qdiscs)

[nddq]

I've looked into the issue and found that eBPF maps are deleted when their Close() function is called, which occurs in the plugin's Stop() function during a graceful shutdown. However, those maps persist if the agent is forcibly terminated via signals like SIGTERM or SIGKILL, bypassing the plugins' Stop() function. To address this, we should implement a goroutine in the main thread to catch these signals and handle cleanup.

[timraymond]

@nddq I think making it crash-only would be more robust (as much as I like defer). On-boot we should check if those maps erroneously exist, then delete and recreate them.