I'm in the process of testing QoS functionality and I really like the "qos statistics performance show" & "qos statistics workload performance show" commands for the data they provide.
However, the output truncates the object names (either the policy group or object name) after 18 characters and I don't see any option for the qos statistics command that will permit me to specify the policy group or workload (object).
I'm concerned about this because in our environment it's possible to have volume names that aren't unique until after the 18th character. How would I know what object I'm looking at if I can't specify the group or object names against the command?
This question is broken into two parts for response.
I. QoS statistics workload's 18 character output restriction
Workaround summary: In Data ONTAP 8.2.x version, two step workflow with "qos workload performance show" and diagnostics privilege "qos workload show" commands can be used to identify the workload ID of the object and map to the object uniquely using the identified workload ID.
(1) Use "qos statistics workload performance show" to list the top 20 workloads. Use the ID column's value for identifying the workload's unique ID.
clus-sparab2::*> qos statistics workload performance show
8195 pg1 vs0 vs0_vol1232443545466565 <== complete name of the workload
- - -
II. 20 row limitation of qos statistics CLI.
There is no way to adjust this. The "qos statistics" UI was designed to be a quick way to show the top-20 workloads that are impacting the system. For regular monitoring of all the
work on a cluster, a better approach is to write a script to send ZAPIs to collect all the performance data. The data shown by "qos statistics" isn't directly available, but with some math, the same performance characteristics can be calculated.