Cluster: IOPS in B/s unit

[matejzero]

Describe the bug

In cluster dashboard, IOPS are shown as Bytes/s unit.

Environment

Harvest version: harvest version 21.05.2-1 (commit ce091de) (build date 2021-06-14T20:31:09+0530) linux/amd64
Command line arguments used: /opt/harvest/bin/harvest restart --config /opt/harvest/harvest.yml
OS: CentOS 7
Install method: yum
ONTAP Version: 9.7
Other:

To Reproduce

Open Cluster dashboard and look at IOPs graph.

Expected behavior

Unit should be iops/s or none.

Actual behavior

Possible solution, workaround, fix

Change unit to IO ops/sec

[cgrinds]

@Hardikl can you take a look? - this is similar to #186

[Hardikl]

yes, it is same as #186, just it was targeted for node dashboard.
Found few other graphs in other dashboard also shows wrong unit. I will cover throughput, iops and latency related graphs in all dashboards and handle them here.

[ruanruijuan]

Thank you Hardik for being proactive on examining these counters on all dashboards!

[sridevimm]

Tested this scenario in Harvest point release 21.05.3.1 and able to reproduce this issue. Cluster IOPs chart still shows unit as kB/s instead of iops/s. Reopening the issue to get this fixed.

[sridevimm]

Node title shows K io/s. But, the charts are showing millisecs and microsecs. Please take a look at the units for those charts as well.

[rahulguptajss]

@Hardikl do check once if the test setup had latest grafana dashboards

[Hardikl]

I have installed the latest binary in RPM[10.140.132.80 root/netapp], alongwith the grafana(8.0.3) and prometheous(2.28.0).
Then imported the dashboards via this:
$ bin/grafana import --addr my.grafana.server:3000 --directory grafana/prometheus

Cluster and node dashboard graphs are showing as expected. Also I have checked that the changes are available in release branch as well, so moving this to testme.

[Hardikl]

@rahulguptajss out of 2 rhel harvest qa deployment, i was not able to access the grafana UI, I am not sure that test setup has latest grafana dashboards.