We have a 3 hosted-engine nodes setup using 2 NFS3 shares on which the engine keeps crashing every few days.

Looking at VDSM logs, it looks like a storage problem but I'm wondering why don't they restart the engine?