It happens when deleting all pods of a tapp with immutable policy.
Resync queries all ips firstly without holding a big lock and then it
tries to hold a pod's lock to check status. But the lock is hold by
the binding routine of the new pod, the binding routine updated the
old pod uid to new pod uid and released the lock. Then resync holds
the lock with the old pod uid attribute and finds it missmatches new
pod's uid and calls unassign ip making the ip unreachable.
It happens when deleting all pods of a tapp with immutable policy. Resync queries all ips firstly without holding a big lock and then it tries to hold a pod's lock to check status. But the lock is hold by the binding routine of the new pod, the binding routine updated the old pod uid to new pod uid and released the lock. Then resync holds the lock with the old pod uid attribute and finds it missmatches new pod's uid and calls unassign ip making the ip unreachable.