Dead balls NFS

Question

NFS

Dear colleagues, tell me this question.

Here I mount a remote NFS server on the NFS client. Let's say, for some reason, after some time, the NFS server picks up and goes out forever. The share on the client was not previously unmounted.

Accordingly, when trying to stat'a (ls for example) a mounted dir, the process that did it is blocked for life by the kernel, which is trying to restore the connection to the dead ball. And no kill's have any effect on him.

Moreover, when soft rebooting this server with an NFS client, the reboot process hangs itself on an attempt to unmount dead shares. And only hard reset helps.

The question is how to make the NFS client kill a dead share on some timeout, and not keep it in the lock for life?