Monitoring a NetApp storage system... something is wrong?

beardmann · March 24, 2026, 2:45pm

CMK version: 2.4.0p24
OS version: Ubuntu 24.04

Error message:

Well where to start

We are trying to setup monitoring of a “simple” NetApp C800.

First challenge:

The aggr0 system agregates are typically quite small, so we mostly ignore the space limits on those… it look me awhile to find out that I needed to use the “filesystem” rule to try to set the limits on the specific aggregates… and I used the “mountpoint” value and added “aggr0*” which seemed to limit this… so far so good… not sure why “filesystems” is used here when there is a rule for NetApp volumes??

Next issue… we create clones of our LUNs and volumes from time to time… and by chance two clones of the same LUN was present when setting up…

This isn’t handled that great… as it shows one LUN (which points to one of the clones) but it doesn’t show any of the other LUNs…

Below are a list of LUNs from the NetApp… the first one is the “main” production LUN, while the two others are clones…

vojstor_vmware
          /vol/ds2/ds2                    online  mapped   vmware    50.40TB
vojstor_vmware
          /vol/ds203102608020891011/ds2   online  mapped   vmware    50.40TB
vojstor_vmware
          /vol/ds203172614573579614/ds2   online  mapped   vmware    50.40TB

For some reason only the last LUN is shown in the list of LUNs… ??

But in fact I would very much like to exclude clones all together, which I cannot find as any of the options in the LUN rule?

I guess this would be the same issue with volumes and flexclones etc.. ?

Then we have the interfaces… as you can see below we have some interfaces which report OK and have traffic, while other reports UNKN and have no traffic..

The ones which are OK are the management LIFs, cluster and intercluster LIFs. While the UNKN LIFs are used for iSCSI… we tag a VLAN on to them, but that’s all… so not really sure why the Check_MK agent doesn’t like them as they are perfectly OK and we are pushing data through them…

I tried to see if there were any thing to see in the “interface rule”, but sadly nothing useful…

There are no traffic details on the physical ports however?

So for now, the check_mk agent for NetApp is just great if you want to see how fast your fans spin… but that’s about it ? Not sure if there are differenced from the “Raw” version and the paid variants? Please let me know because this is a test to see if we should switch to the paid version… right now I have to say it doesn’t look that promising, for us at least.. maybe someone else will have better luck…

If someone can guide me a bit, that would be great

Yggy · March 24, 2026, 3:06pm

I’m clueless about LUNs, but read something similar before on the forum, so searched and found following:

Regarding interfaces, you can rename them by description or alias, so its easier to manage or even disable some that have little to no use. See following blog article how to rename interfaces:

r.sander · March 24, 2026, 3:38pm

The UNKNOWN for the interface checks comes form the fail-over-policy which Checkmk seems to not know.

github.com/Checkmk/checkmk

packages/cmk-plugins/cmk/plugins/netapp/agent_based/netapp_ontap_if.py

master


      
          if interface.get("port_speed"):
              computed_speed = interface["port_speed"] * 1000 * 1000
          
          interface["state"] = computed_state
          interface["speed"] = computed_speed
          
          if interface.get("mac-address"):
              if_mac_list.setdefault(interface["mac-address"], [])
              if_mac_list[interface["mac-address"]].append((interface_name, computed_state))
          
          failover_policy = interface.get("failover")
          fail_ports: set = broadcast_domains_ports.get(interface.get("broadcast_domain", ""), set())
          # see FAILOVER_REST_TRANSLATION in lib.py for mapping rest-cli failover names
          match failover_policy:
              case "default" | "broadcast_domain_only":
                  interface["failover_ports"] = ";".join(el for el in fail_ports)
              case "home_node_only":
                  interface["failover_ports"] = ";".join(
                      el for el in fail_ports if el.split("|")[0] == interface["home-node"]
                  )
              case "home_port_only":

it knows about “default”, “home_node_only” and “home_port_only”.

“sfo-partner-only” has to be added in the code.

r.sander · March 24, 2026, 3:43pm

The NetApp special agent produces service checks that start with “Interface”, but doe snot use the if64 check plugin. There is a netapp_ontap_if check plugin which currently does not implement interface service renaming AFAIK.

There is only a “Netapp port discovery” ruleset with which you can ignore certain interface types.

Yggy · March 24, 2026, 3:51pm

Silly me thinking all interfaces were created the same way as service.
I feel even more clueless now …

beardmann · March 24, 2026, 3:59pm

Might be… but why isn’t it showing any data via these LIFs ? … or via the physical ports? Is that normal? you basically has no idea of what the interfaces are doing… only the management and intercluster interfaces have traffic details… must be a bug?