uneven shale
Hey all, I just updated but it seems like my inter-cluster connections are not still showing actual usage... Here's a screenshot of the NetApp GUI showing throughput on the e1X and e7X ports (switchless config, just direct connect between the controllers)
But NAbox is showing almost zero traffic on the interfaces at the same time...
Harvest 25.11.0
NABox 4.1.0
I grabbed a support bundle at this time, but I'm not sure where the hiccup is
proven solar
@uneven shale Could you check time series graph below if data over a time range makes sense. Table has instant data so it may differ depending on polling cycle of Harvest and System manager.
You can also run below command in diag mode in ontap cli to compare harvest nic metrics vs ontap.
set d
statistics show-periodic -interval 60 -iterations 50 -object nic_common -instance e0a -counter tx_bytes|rx_bytes
Thanks @proven solar, it looks like the GUI is pulling vastly different info from something... The GUI gets updated about every 15 sec and is constantly in the three digit MB/s numbers... but the command you're showing (and grafana) don't come anywhere near that...
proven solar
Do you think the values shown by the system manager are correct, based on the traffic you expect from these ports? The system manager invokes a different endpoint to collect this information.
api/network/ethernet/ports?fields=*,statistics
Harvest invokes api/cluster/counter/tables/nic_common
uneven shale
Do you think the values shown by the system manager are correct, based on the tr...
We're hammering this system (it's a demo unit and we're stress testing the iSCSI connections, and we're limited on iSCSI ports as part of the test.) I don't have enough experience with the NetApp to know if it should be sending data through those ports or not with this workload... I can't tell if MPIO would be spreading the workload across both controllers instead of going direct through the local ports per workload
proven solar
Ok. We'll check internally against these endpoints.
steel mica
@uneven shale We have checked locally, as Rahul mentioned SM consumes api/network/ethernet/ports api with polling interval as 15s, Harvest would consumes api/cluster/counter/tables/nic_common api with polling interval as 1m. I have tried to change polling interval of this template to 15s (same as SM). Values seems to be matching, attached the screenshot of the same.
uneven shale
Thank you @steel mica, I'll look and see if the polling interval change helps with what I'm seeing. I wanted to bring it up, because I thought it was a bug that you guys should be aware of, but if it's not an issue, then I'll try to recreate the steps and see if I can get the info I'm looking for.
Thanks for being so responsive!