Why is there a discrepancy in CPU usage between the results of 'qos statistics volume resource cpu show' command and 'sysstat' command in node shell?
Applies to
- ONTAP 9
- While checking the CPU usage by using qos statistics volume resource cpu show command and sysstat command in node shell, the results are as follows:
18:12:31::> qos statistics volume resource cpu show -node node1
18:12:39 Workload ID CPU
18:12:39 --------------- ----- -----
18:12:39 -total- (100%) - 0%
18:12:41 -total- (100%) - 0%
18:12:42 -total- (100%) - 0%
18:12:43 -total- (100%) - 0%
18:12:44 -total- (100%) - 0%
18:12:45 -total- (100%) - 0%
18:12:46 -total- (100%) - 0%
18:12:47 -total- (100%) - 0%
18:12:48 -total- (100%) - 0%
18:12:49 -total- (100%) - 0%
18:12:50 -total- (100%) - 0%
18:12:51 -total- (100%) - 0%
18:12:52 -total- (100%) - 0%
18:12:53 -total- (100%) - 0%
18:12:54 -total- (100%) - 0%
18:12:55 -total- (100%) - 0%
18:12:56 -total- (100%) - 0%
18:12:57 -total- (100%) - 0%
18:12:58 -total- (100%) - 0%
18:12:59 -total- (100%) - 0%
18:13:00 -total- (100%) - 0%
::> system node run -node node1 -command sysstat -x -c 900 5
18:12:30 CPU
18:12:30 92%
18:12:35 94%
18:12:40 96%
18:12:45 95%
18:12:50 96%
18:12:55 97%
18:13:00 96%
18:13:05 89%
Answer
qos statistics volume resource cpu showcommand measures the CPU usage based on QoS volumes, whereassysstatcommand in the node shell measures the total CPU usage of the entire node, regardless of the presence of volumes.- In the command outputs above, since there is no real IO workload on the QoS volumes, the total CPU usage for the entire node, based on both QoS volumes and their workloads, is also 0%.
Note: When the volume is created, the QoS workload ID is automatically assigned to ONTAP.
