So you just completed a round of firmware updates (everyone’s favorite!), and all of your hosts have been rebooted. Within a few days your customers are complaining about poor performance on their VMs. It has to be something on their side, right? Well, not necessarily. There is a bug in ESXi 6.5, and 6.7 where NFS Performance Decreases within a few days after a Host Reboot.

According to the KB from VMware, symptoms of the bug include the following:

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

At the time of this writing, there is no resolution for the bug. The workaround is to increase the following host parameter to 32 or 64 (default is 4):

esxcfg-advcfg -s 32 /SunRPC/MaxConnPerIP

There is no impact to implementing this workaround. However, it does require a reboot.