wanton crypt Sep 3, 2024, 4:35 AM

#

Hi ALl, from the harvest document,
I found this metric: https://netapp.github.io/harvest/23.02/ontap-metrics/#svm_cifs_connections
It can show the current numbers of connection of cifs

But I cannot find related clients counts for NFS and iscsi,
Can anyone help with this.
We want to monitor this number and compare to the limit

hushed walrus Sep 3, 2024, 5:54 AM

#

@wanton crypt See if below templates help

name:             NFSClients
query:            api/protocols/nfs/connected-clients
object:           nfs_clients

counters:
  - ^^client_ip                    => client_ip
  - ^^node.name                    => node
  - ^^protocol                     => protocol
  - ^^server_ip                    => server_ip
  - ^^svm.name                     => svm
  - ^^volume.name                  => volume

plugins:
  - LabelAgent:
      value_to_num_regex:
        - count protocol .* .* `0`
  - Aggregator:
      # plugin will create summary/average for each object
      # any names after the object names will be treated as
      # label names that will be added to instances
      - node
      - svm
      - volume

# only export node/aggr aggregations from plugin
# set this true or comment, to get data for each lock
export_data: false

name:             ISCSISessions
query:            api/protocols/san/iscsi/sessions
object:           iscsi_sessions

counters:
  - ^^tsih                         => tsih
  - ^^target_portal_group          => target_portal_group
  - ^^initiator.name               => initiator
  - ^^isid                         => isid
  - ^^svm.name                     => svm

plugins:
  - LabelAgent:
      value_to_num_regex:
        - count initiator .* .* `0`
  - Aggregator:
      # plugin will create summary/average for each object
      # any names after the object names will be treated as
      # label names that will be added to instances
      - svm

# only export node/aggr aggregations from plugin
# set this true or comment, to get data for each lock
export_data: false

Below queries shall get you the total count.

wanton crypt Sep 3, 2024, 6:09 AM

#

Do we support these metrics for ZAPI?

hushed walrus Sep 3, 2024, 6:12 AM

#

What is your ONTAP version?

#

These Rest Points are available from 9.6/9.7 versions. For iscsi, i see that iscsi-session-get-iter is available.

wanton crypt Sep 3, 2024, 9:27 AM

#

@hushed walrus 9.11 and above
Yes, I mean that do they support in ZAPI template?
Or only in REST?

hushed walrus Sep 3, 2024, 9:55 AM

#

The templates I shared above are for REST. If you are using clusters with version 9.7 or higher, you can use these templates. For ZAPI, you can create a template for ISCSI using the iscsi-session-get-iter ZAPI. For NFS, see if the nfs-service-get-iter ZAPI is helpful.

wanton crypt Sep 3, 2024, 9:57 AM

#

So I have to create two templates under /zapi/cdot/9.8.0/, one for iscsi_session.yaml, and another is nfs_service.yaml?

hushed walrus Sep 3, 2024, 10:01 AM

#

Yes

#

Pls do check if NFS zapi is giving you the results as you expect.

wanton crypt Sep 10, 2024, 7:07 AM

#

@hushed walrus I found this metric: https://netapp.github.io/harvest/23.02/ontap-metrics/#node_cifs_connections
node_cifs_connections.
And user want to monitor the counts of the connection of a node. From the HWU, the limit of this value match to which of the two following?

hushed walrus Sep 10, 2024, 7:09 AM

#

Could you share links of HWU?

wanton crypt Sep 10, 2024, 7:12 AM

#

hushed walrus Could you share links of HWU?

https://hwu.netapp.com/Controller/Index?platformTypeId=30310846&platform=29266985&os=34942206#none

hushed walrus Sep 10, 2024, 7:20 AM

#

It should be

#

You can double check from ONTAP team once.

wanton crypt Sep 10, 2024, 7:25 AM

#

@hushed walrus From the description, it says "Maximum number of NAS (SMB and NFS) TCP connections that a node can support", but isn't this metric only for CIFS?

hushed walrus Sep 10, 2024, 7:26 AM

#

Yes, This is for CIFS only.

wanton crypt Sep 10, 2024, 7:28 AM

#

But how does this match the maximum number? (which says SMB and NFS)
I am afraid that we set the wrong threshold using this metric to monitor the connection of a node

#

Does that mean that we have to add this CIFS metric with the NFS metric for NFS connected clients counts, which I had opened a feature request and compare the sum to this limitation?

hushed walrus Sep 10, 2024, 7:33 AM

#

The ZapiPerf template for SMB2 includes a metric called connected_shares. Currently, Harvest does not collect this metric. Also, the RestPerf does not have this metric at all.

#

It will probably be cifs + smb2 + nfs

wanton crypt Sep 10, 2024, 7:36 AM

#

which is TCP connections
Are we able to create a metric for this value?

hushed walrus Sep 10, 2024, 7:40 AM

#

Didn't get your question. Metric for which value?

wanton crypt Sep 10, 2024, 7:41 AM

#

Metric for the "maximum number of connections - NAS" value from HWU,
Which shows 100,000.

hushed walrus Sep 10, 2024, 7:42 AM

#

Sure. Please open a feature request for the same.

hushed walrus Sep 19, 2024, 7:40 AM

#

@wanton crypt I have updated issue with relevant steps
https://github.com/NetApp/harvest/issues/3136#issuecomment-2360217955
Please check and let us know if it works for you. Thanks.

wanton crypt Sep 19, 2024, 7:43 AM

#

Is it possible to support ZAPI?

hushed walrus Sep 19, 2024, 7:45 AM

#

Could you provide cluster version?

wanton crypt Sep 19, 2024, 8:23 AM

#

9.11.1

hushed walrus Sep 19, 2024, 8:26 AM

#

The REST endpoint api/protocols/nfs/connected-clients is available from ONTAP version 9.7 onwards. Similarly, the REST endpoint api/protocols/san/iscsi/sessions is available from ONTAP version 9.6 onwards.

Therefore, if your clusters are running these versions or later, collecting data via REST should work fine for you.

For iSCSI, there is a ZAPI mapping available with the iscsi-session-get-iter call. However, I couldn't find a relevant ZAPI call for NFS clients.

Please note that the NFS clients API provides data for clients active in the last 48 hours. If you need real-time data, you will need to add a filter for idle_duration in the template.

wanton crypt Nov 14, 2024, 1:21 AM

#

@hushed walrus Here is the update.
We tested the nfs connection metric
And it matches the CLI: "vserver nfs connected-clients show",
But the node limits should be using the CLI: "network connections active show -protocol nfs -node XXX"

I think there is a gap between this two value

Or if adding idle_duration filter to the template, they will be the same value?

hushed walrus Nov 14, 2024, 6:36 AM

#

I think so. More details about these are here
https://docs.netapp.com/us-en/ontap-cli/network-connections-active-show.html#description
https://docs.netapp.com/us-en/ontap-cli/vserver-nfs-connected-clients-show.html

I think network connections active show is refreshing data every 30 seconds and it is likley that it only shows active connections whole vserver nfs connected-clients show displays last 48h connections.

wanton crypt Nov 14, 2024, 1:25 PM

#

So is it is possible to create a template for this connected-clients right? Using network connections active show

hushed walrus Nov 14, 2024, 1:26 PM

#

Yes

wanton crypt Nov 15, 2024, 7:08 AM

#

@hushed walrus I cannot find related API for this command

#

“network connections active show”

hushed walrus Nov 15, 2024, 7:11 AM

#

Have you tried api/private/cli/network/connections/active ?

wanton crypt Nov 15, 2024, 7:12 AM

#

hushed walrus Have you tried `api/private/cli/network/connections/active` ?

This works, I'll try to create the template

hushed walrus Nov 15, 2024, 7:12 AM

#

Sure, Let us know if any issues.

wanton crypt Nov 15, 2024, 7:47 AM

#

Is this correct?
With "?service=nfs" after the query

hushed walrus Nov 15, 2024, 7:47 AM

#

Could you share template in text format?

wanton crypt Nov 15, 2024, 7:47 AM

#

`name: NFSClientsConnection
query: api/private/cli/network/connections/active?service=nfs
object: nfs_clients

counters:

^^node => node
^^vserver => svm

plugins:

LabelAgent:
value_to_num_regex:
- count protocol .* .* 0
Aggregator:
plugin will create summary/average for each object

any names after the object names will be treated as

label names that will be added to instances
- node
- svm

- volume

only export node/aggr aggregations from plugin

set this true or comment, to get data for each lock

export_data: false`

hushed walrus Nov 15, 2024, 7:47 AM

#

Thanks. Let me check.

#

Can you try this

name:             NFSClientsConnection
query:            api/private/cli/network/connections/active
object:           nfs_clients

counters:
  - ^^cid
  - ^node                       => node
  - ^vserver                    => svm
  - filter:
      - service=nfs

plugins:
  - LabelAgent:
      value_to_num_regex:
        - count cid .* .* 0
  - Aggregator:
      - node

# only export node/aggr aggregations from plugin
# set this true or comment, to get data for each lock
export_data: false

wanton crypt Nov 18, 2024, 3:47 AM

#

@hushed walrus
Seems like in older version of ONTAP, ex:9.7,
The number is different.

And in version before 9.5, do we support ZAPI for this?

#

1842 vs 375 counts in ONTAP 9.7

hushed walrus Nov 18, 2024, 6:31 AM

#

Okay. Do you see any duplicate warnings in Harvest logs?
Could you email the output of below curl @ ng-harvest-files@netapp.com

curl -s -k -u USER:PASS 'https://CLUSTER_IP/api/private/cli/network/connections/active?service=nfs'

wanton crypt Nov 18, 2024, 7:24 AM

#

The curl output is the same as the metric

hushed walrus Nov 18, 2024, 7:25 AM

#

Could you email the output

wanton crypt Nov 18, 2024, 7:42 AM

#

I can only took the screenshot

hushed walrus Nov 18, 2024, 7:42 AM

#

Thanks. Seems like Harvest number is matching then?

wanton crypt Nov 18, 2024, 7:43 AM

#

Yes,
Seems like API number is not the same as the CLI number

hushed walrus Nov 18, 2024, 7:44 AM

#

Possibily related to 9.7 version. Seems correct in 9.15

wanton crypt Nov 18, 2024, 7:44 AM

#

9.9 and above version are correct

#protocol connection total metrics

plugin will create summary/average for each object

any names after the object names will be treated as

label names that will be added to instances

- svm

- volume

only export node/aggr aggregations from plugin

set this true or comment, to get data for each lock