#Dashboard ONTAP Network Receive Errors

1 messages · Page 1 of 1 (latest)

old ember
#

Say ... I noticed some spikes in this graph for one of our data centres. But the table under the graph looks weird to me. I expected to see a list of cluster nodes and interfaces i.e. where the errors are have occurred. But instead I see a series of rows all named TOTAL or ALIGMENT ... I.e. The type of the error, but not the source.

#

Hovering over a peak on the graph results in a similar issue : a hover-box/popup shows me: "11:11:40 green TOTAL 54"

#

Also, and I'm not sure if it means anything, the color coding seems wrong. I.e. when I get above popup with the green swatch, I am actually hovering over an orange plot ...

charred gust
#

Hi @old ember Yes, I see what you mean. I agree this is not very useful. That panel is meant to show errors by cluster and then the Ethernet port errors panel below has the cluster, node, nic breakdown. Does the Ethernet port errors panel have what you want? We should change the NICs Receive Errors by Cluster panel to include the cluster though like so

image.png image.png
old ember
#

Right. The panel below shows me a nice detail table with: "DC ; Cluster ; Nic ; Node and counts for the various error types"

#

So I am good with that one (although the Node and "Nic" columns could be swapped ;-))

charred gust
#

like so?

image.png
old ember
#

That order makes more sense to me ...

#

Also, I wonder a bit about the terminology of "Port" and "NIC "... Do they have different meanings, or could they be rationalised?

charred gust
#

agreed - least to most specific

#

we can't really change the exported metric label but we can change that dashboard to say port instead of nic. Would that work for you?

old ember
#

It would seem more consistent, unless there is something I am missing ...

#

Everyone is a UI design critic 🤓

charred gust
#

I don't think you're missing anything, but we'll double check and make that change if it makes sense

old ember
charred gust
#

excellent! thanks for the feedback

old ember
#

No worries!

old ember
#

Humm ... the second panel of that dashboard is titled "Ethernet" and it contains two tables: one entitled "Ethernet ports" and the other "NIC ports" ... so some difference is implied. What I see here, listed in "Ethernet ports" is a mixture of physical interfaces e.g. "e1c" and what I would call LIFs i.e. ONTAP Logical Interfaces with names like "a0a-1234". The second table seems to list only physical Ports like "e0M" or "e10c". At least some interfaces appear in both tables 😕 ... The first table seems to emphasis configuration info. and the second has more throughput info.

#

In table "Ethernet ports" the MTU values could be better displayed "raw" e.g. as 1514 rather than 1.46KiB

charred gust
#

ah yes, good catch

old ember
#

Of course, maybe not all NICs / Ports are necessarily Ethernet (they are here) ...

#

Glad to help 🤪

charred gust
#

right - the Ethernet ports table shows metrics from the ONTAP endpoint api/network/ethernet/ports while the NIC ports table shows metrics from the ONTAP endpoint api/cluster/counter/tables/nic_common

old ember
#

Makes sense. Personally I'm OK wih Physical Interface == NIC == Port. I'm not so sure about that first table containing Ethernet and Port together ... maybe "Ethernet LIF" or "Ethernet Interface" would be more "ontappy". Glad to have the Dashboard and the data though!

charred gust
#

You mean that the row name should be Ethernet LIF or Ethernet Interface instead of Ethernet or are you suggesting a change to the table names?

image.png
old ember
#

I checked the swagger interface docs and it describes "api/network/ethernet/ports" (the basis of that first table) as returning "a collection of ports (physical, VLAN and LAG) for an entire cluster ..." So that table name should remain as is.

#

For the second table ONTAP itself returns (curl) the description "This table tracks hardware network traffic performance and errors for all supported network interface cards (NIC), such as Intel Niantic, Qlogic, and CNA." ... So maybe that one could be re-labelled as "Port perfomance" or "Port statistics". I don't personally feel that the "NIC" nomenclature really brings any added value (except that it is consistent with the API call) ... But IANAOD (I Am Not An ONTAP Designer)

charred gust
#

Thanks @old ember that's a good suggestion, we will incorporate. IANAOD either 🙂

charred gust
#

@old ember most of these are incorporated in the 25.05.1 build. That build also includes the ONTAP: Switch dashboard that we discussed in a different thread

old ember
#

Also, I notice a fair number of receive drops on the (virtual) Ethernet interface of the nabox guest (e.g. using: ifstat-iproute2 -ejp ens192). So I'd like to enable jitter to try to spread the traffic load and see if that makes any difference. Any tips on how to do that on nabox?

charred gust
#

NAbox4 does not currently support this cleanly. To change jitter in Nabox4, you need to modify Harvest's out-of-the-box default.yaml, e.g. sudo vim /data/packages/harvest/conf/rest/default.yaml
That is unfortunate because every time you upgrade Harvest, that file will be overwritten.

I've included a screenshot of what you want to add. You can add jitter to all the templates you use: Rest /data/packages/harvest/conf/rest/default.yaml, RestPerf /data/packages/harvest/conf/restperf/default.yaml, KeyPerf /data/packages/harvest/conf/keyperf/default.yaml, etc.

image.png