NetApp
Limit which instances are collected
Discussed in https://github.com/NetApp/harvest/discussions/867
Originally posted by luddite516 February 17, 2022 We have a use case where we are collecting file lock statistics for a specific vserver. I have built the following template to collect lock stats:
name: Locks query: lock-get-iter object: lock
collect_only_labels: true
counters: lock-info: - ^node - ^volume - ^vserver - ^path - ^^lockid - ^client-address => client_ip - ^smb-open-type => smb_open_type
export_options: instance_keys: - lockid instance_labels: - vserver - volume - node - path - client_ip - smb_open_type
The above template collects all the locks in the cluster, and there may be hundreds of thousands of them.
Is there any way to limit which instances are collected, to reduce load on cluster? For example, limit collection to only a specific vserver or volume?
Thanks.
@luddite516 as of Harvest 24.02, filtering is supported by the Rest and ZapiPerf collectors. We still don't have filtering for Zapi, would moving to Rest work you?
ONTAP added api/protocols/locks in ONTAP 9.10.0, REST private CLI for 9.6+ may also work for you if you're unable to upgrade.
Yes, I believe we could move to REST. We are going to have to get there eventually.
On Wed, Feb 28, 2024 at 9:10 AM Chris Grindstaff @.***> wrote:
@luddite516 https://github.com/luddite516 as of Harvest 24.02, filtering is supported by the Rest and ZapiPerf collectors. We still don't have filtering for Zapi, would moving to Rest work you?
ONTAP added api/protocols/locks in ONTAP 9.10.0, REST private CLI for 9.6+ may also work for you if you're unable to upgrade.
— Reply to this email directly, view it on GitHub https://github.com/NetApp/harvest/issues/868#issuecomment-1964879359, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKO2LMWI4HLIVG7IEK45SYTYV43E7AVCNFSM5OVRZM3KU5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TCOJWGQ4DOOJTGU4Q . You are receiving this because you were mentioned.Message ID: @.***>
@luddite516 See if the lock templates for Zapi and Rest are helpful. These templates only provide the count of locks at the node, svm, lif, and volume levels. They are designed to avoid collecting thousands of lock instances, as requested in this issue.
