Questions about embedshim reliability based on sanbox API

[silence-coding]

I have a question. In the shimv2 scenario, if a shim process is used to manage all containers, even if the shim process takes over the original container through ebpf after abnormal restart, all containers will be stopped because containrd invokes the stop command of shim. In the sanbox API scenario, if embedshim is an independent process and exits abnormally, will containrd forcibly delete (restart) the container managed by embedshim?