The VM NUMA CPU assignments are sampled too often causing the load average on hypervisor to increase over time.

Issue

• The load average on the hypervisor grows over the time.

• Hypervisor can also transition to Non-Operational state.

• VM running on the hypervisor change to pause state.

• Restarting the vdsmd daemon process decreases the load average but starts to increase over time.

 00:21:43 up 68 days,  4:42,  1 user,  load average: 30.11, 25.16, 22.70

#systemctl stop vdsmd

 00:22:33 up 68 days,  4:42,  2 users,  load average: 29.40, 25.83, 23.06
 00:22:53 up 68 days,  4:43,  2 users,  load average: 26.20, 25.33, 22.95
 00:23:01 up 68 days,  4:43,  2 users,  load average: 24.32, 24.95, 22.86
 00:23:36 up 68 days,  4:43,  2 users,  load average: 16.58, 22.95, 22.26
 00:24:11 up 68 days,  4:44,  2 users,  load average: 12.09, 21.13, 21.68
 00:24:24 up 68 days,  4:44,  2 users,  load average: 11.06, 20.61, 21.50
 00:24:38 up 68 days,  4:44,  2 users,  load average: 9.49, 19.80, 21.22
 00:24:53 up 68 days,  4:45,  2 users,  load average: 7.98, 18.96, 20.93

#systemctl start vdsmd

 00:26:00 up 68 days,  4:46,  2 users,  load average: 5.69, 16.23, 19.86
 00:25:15 up 68 days,  4:45,  2 users,  load average: 8.13, 18.29, 20.67
 00:25:44 up 68 days,  4:46,  2 users,  load average: 6.35, 16.89, 20.13
 00:26:12 up 68 days,  4:46,  2 users,  load average: 5.87, 15.76, 19.65
 00:26:27 up 68 days,  4:46,  2 users,  load average: 5.60, 15.22, 19.41
 00:26:53 up 68 days,  4:47,  2 users,  load average: 5.55, 14.44, 19.04
 00:27:26 up 68 days,  4:47,  2 users,  load average: 6.61, 13.83, 18.69
 00:27:58 up 68 days,  4:48,  2 users,  load average: 7.31, 13.17, 18.29
 00:28:43 up 68 days,  4:49,  2 users,  load average: 15.40, 14.34, 18.43
 00:29:53 up 68 days,  4:50,  2 users,  load average: 20.66, 16.20, 18.78
 00:30:55 up 68 days,  4:51,  2 users,  load average: 25.88, 18.48, 19.40
 00:31:40 up 68 days,  4:51,  2 users,  load average: 28.35, 20.13, 19.92