VMs network disconnect every hour after installing VC Patch 2

[berezuk]

Prior to installing VC patch 2 everything was running fine. Now after installing patch 2, all of my VMs on all of my clusters and ESX 3.0.1 hosts just disappear from the network for a minute or two every hour at exactly the same time of the hour. Looking at the performance statistics, there is a drop to zero percent of CPU at the time the VM disappears and disconnects from users. The only common log entry is in the ESX host hostd.log with the entry:

2007-09-26 14:01:00.722 "Statssvc' 104561584 verbose Propagating stats from interval 20 to 300

...It almost seems as if the patch 2 statistics roll-up fix is causing the VMs to go to sleep when their host is reporting their statistics.

I have not experienced this on VC patch 1 or VC 2.0.2 .

[VirtualNoitall]

Hello,

Is your service console not getting the CPU it needs ( high wait times ) and/or does the SC have enough available memory?

[berezuk]

Its been consistent with ESX hosts heavily loaded vs almost totally free. In clusters and non-clustered hosts. I cannot tell if the service console itself lacks CPU or memory. It is not the service console connection that breaks, its the VM's VLAN connection that breaks. VC remote control still is operational.

In one particular instance I have only 2 win2003 VMs running on a single 32GB ESX host and every hour, they loose their network connection or have died. After a minute or so they are back. Using the VI client remote control seems to always get in and may actually trigger it to get 'reloaded' into memory.