Recently, I found that many processes in our server experienced high frequency of major page faults. The disk IO on this server is also very high. I also noticed that when I execute "top" command, for some processes, the value in SHR column became zero. when I restart these processes, the values of SHR became non-zero again. What could be the reason for frequent major page fault? Is there any connection between zero SHR and major page fault?