NFS datastores decrease within a couple of days after host reboot

1. The performance of VMs decrease on NFS datastores within a couple of days after a host reboot.
2. IO Meter tests show increasing latencies and a significant drop of IOPS after a week, and gets worse after 2 to 3 weeks.
3. A host reboot fixes the issue temporarily.

Login via ssh.

[root@Dtech-ESXi-01:~] esxcfg-advcfg -g /SunRPC/MaxConnPerIP
Value of MaxConnPerIP is 4

[root@Dtech-ESXi-01:~] esxcfg-advcfg -s 32 /SunRPC/MaxConnPerIP
Value of MaxConnPerIP is 32

The value should be bigger than the maximum number of shares mounted from the same IP address.
Reboot the host after changing the value.

Note: There is no impact by implementing the above workaround.