Trying to enable NFS performance stats

[70tas]

I see there are NFS settings being discussed, but I can't figure out how to enable their collection. Can anyone help me set up the configuration for NFS v3/v4?
I think I need to copy the /data/packages/harvest/conf/zapiperf/default.yaml to /etc/nabox/harvest/user/restperf/... cdot???, but that is where I am stuck.

[rahulguptajss]

@70tas These templates are already enabled by default for all collectors (ZapiPerf, RestPerf, KeyPerf). You shouldn’t need to do anything to enable them.

Are you not getting these metrics? Their prefix is svm_nfs, and the templates for RestPerf look like the examples below. If these metrics are mising, Please share Harvest logs @ https://upload.nabox.org/pupy-jece-weho .

https://github.com/NetApp/harvest/blob/main/conf/restperf/9.12.0/nfsv3.yaml
https://github.com/NetApp/harvest/blob/main/conf/restperf/9.12.0/nfsv4.yaml
https://github.com/NetApp/harvest/blob/main/conf/restperf/9.12.0/nfsv4_1.yaml
https://github.com/NetApp/harvest/blob/main/conf/restperf/9.12.0/nfsv4_2.yaml

[70tas]

@rahulguptajss
Hi Rahul:
My explorer show both svm_nfs and node_nfs entries. I can only get data from svm_nfs_read_avg_latency and svm_nfs_write_avg_latency. I could only get ops from node_nfs_ops. Is that how it should be?

I am also trying to get back-end numbers, but ChatGPT is suggesting I find 'pollers' in my configs, which I don't have. Are pollers a legitimate option for the configuration files? If so can you point me to some instructions on how to enable them?
Thanks
TasP

[rahulguptajss]

@70tas Could you tell me which NFSv3/v4 metrics you’re looking for? There are several svm_nfs prefixed metrics available in Harvest, as listed in the metric docs: https://netapp.github.io/harvest/26.02/ontap-metrics/#svm_nfs_access_avg_latency

Can you share harvest logs at https://upload.nabox.org/miqy-wiro-saso, and we can check whether the relevant templates are enabled.

[70tas]

Thank you @rahulguptajss . Unfortunately I cannot share logs, we are a dark site with no internet access. I think I've found what I was looking for, however:
I was able to pull latency numbers using svm_nfs_r/w_avg_latency, and NFS IOPs with (node_nfs_ops) / 1000.
Is that the proper way of presenting these numbers?

[rahulguptajss]

@70tas Harvest provides latency is microseconds and for ops it is ops/sec. So if you divide ops by 1000, it becomes Kops/sec.