Adding the teardown spec to the task state for persistence via the client allows recovering a tasks network subsystem correctly in the event of a latter destroy. Without this, the network controller would need to regenerate the assumed iptables rules, which could be wrong due to changes between failure and recovery.
Adding the teardown spec to the task state for persistence via the client allows recovering a tasks network subsystem correctly in the event of a latter destroy. Without this, the network controller would need to regenerate the assumed iptables rules, which could be wrong due to changes between failure and recovery.