Wait a few minutes, then try the command | NetApp | Page 1

solid quail Dec 18, 2024, 7:59 AM

#

Try to install a new C800 but we get some erorrs when we try to create Ifgrp or change card to storage mode (CX5 Card)
(Ontap 9.14.1P8)

storage port modify -node localhost -port e5a -mode storage

internal error. The network management subsystem on node "localhost" has not
yet completed its startup sequence. Wait a few minutes, then try the command
again.

slot 5: Dual 40G/100G Ethernet Controller CX5
e5a MAC Address: 9c:63:c0:13:72:d4 (auto-unknown-fd-down)
e5b MAC Address: 9c:63:c0:13:72:d5 (auto-unknown-fd-down)
Device Type: CX5 PSID(NAP0000001002)
Firmware Version: 16.26.4012
Part Number: 111-04025
Hardware Revision: B3
Serial Number: MT2413400DEG

::> ifgrp create -node localhost -ifgrp a0a -distr-func port -mode multimode_lacp

Error: command failed: Cannot run this command because the system is not fully
initialized. Wait a few minutes, and then try the command again.

Have never seing that before, do we need to create og add node to a cluster before we can change thease settings ?

mental quartz Dec 18, 2024, 8:05 AM

#

I suppose the cluster must be fully functional before you create the ifgrp

modern pendant Dec 18, 2024, 9:50 AM

#

did you correctly cable the HA/Cluster interconnect connections? Is the chassis/controller ha-config set correctly in maint.mode?

#

do a storage failover show and check that each node is able to communicate with its HA partner

frank pasture Dec 18, 2024, 11:02 AM

#

The only time I’ve seen that message is when a single node is configured. Configure the ha partner and the command will then work

mental quartz Dec 19, 2024, 8:27 AM

#

frank pasture The only time I’ve seen that message is when a single node is configured. Config...

For the node name to even be localhost shows that the node is not in a cluster yet

solid quail Dec 19, 2024, 10:48 AM

#

modern pendant do a `storage failover show` and check that each node is able to communicate wit...

I cannot see the what ifgrp should have relationship to do with HA/Cluster interconnect configiration. We have now a dialog with Netapp find a way to get forward the new installation.

modern pendant Dec 19, 2024, 10:58 AM

#

sorry, misunderstanding. You cannot configure ifgroups until your cluster is completely up and running. That's why you need to check why it is not working. Storage failover show will tell you e.g. if the communication through the HA interconnects is not working correctly, which is one reason why the cluster might not be up

#

it's like trying to troubleshoot Windows Firewall when your server doesn't even boot past BIOS 😉

frank pasture Dec 19, 2024, 11:40 AM

#

Just do cluster setup.
You can’t do a lot of things until the node is in a cluster

Changing a port from storage to network or network storage
Create an ifgrp
As a vlan
Create an aggregate

All these items get stashed into the cluster databases. If there is no cluster there is no database

modern pendant Dec 19, 2024, 11:50 AM

#

Changing a port from storage to network or network storage
I suppose this will also work in maint.mode or before cluster creation, because otherwise the cluster might not see its root aggregate, and without root aggregate you can't do cluster setup

frank pasture Dec 19, 2024, 12:24 PM

#

No. You just end up with a single path until it can be flipped

frank pasture Dec 19, 2024, 1:04 PM

#

Even setting up the first node of the cluster, you are then able to manipulate everything on the node. When the second node joins the cluster, the cluster database will be replicated. Good to go.

frank pasture Dec 19, 2024, 1:07 PM

#

solid quail Try to install a new C800 but we get some erorrs when we try to create Ifgrp or ...

Just run cluster setup on this node. If it is the first node, create the cluster. If you need to add it to a cluster, then "join" the cluster. Note with 9.14, depending on how you add the node you will need a username/password and a cluster-ip to join the cluster. Those commands have never worked until the node is in a cluster (I see the ::> prompt which indicates this node is not configured)

frank pasture Dec 19, 2024, 1:38 PM

#

solid quail I cannot see the what ifgrp should have relationship to do with HA/Cluster inter...

What do you need the ifgrp for before the cluster is ready? Like I said that ifgrp is created inside one of the replicated cluster databases. If the database doesn’t exist you can’t create it and you get that message

frank pasture Dec 19, 2024, 1:40 PM

#

solid quail I cannot see the what ifgrp should have relationship to do with HA/Cluster inter...

It’s not the ha/cluster interconnect. It’s the “cluster” itself. If this node is the start of the cluster, set it up and then make your ifgrp. If you need to add it to a cluster, an ifgrp definitely is not needed. Add the node to the cluster and make the ifgrp

solid quail Dec 19, 2024, 1:46 PM

#

Thanks @frank pasture This is an 8 node IP Metrocluster where we want to add 2 * C800 (1 pr site), which should replace the current 8 nodes FAS9000, then tiers cold blocks to (2*8 nodes SG6160).
Have access to the enviroment tomorrow where we will try joining c800 nodes to the existing cluster and see if it makes a difference

frank pasture Dec 19, 2024, 1:48 PM

#

It will make all the difference! I’ve personally tried in the past (I was trying to go faster than technology will allow me to go!) and it fails with the same message. As soon as the node is the first node in a cluster or had joined a cluster, those commands will work with or with out ha/storage-failover enabled

modern pendant Dec 19, 2024, 4:19 PM

#

frank pasture No. You just end up with a single path until it can be flipped

But that doesn't work if you have to switch all your NICs to storage in the first place, there is no single path then 😄 same chicken-egg problem as with FC ports set to target in an MC FC: boot loop. That's why the storage port modify -p <port> -m storage command is available in maintenance mode

frank pasture Dec 19, 2024, 4:22 PM

#

You don’t have to for a new system man. By default either there are onboard disks already configured is there is a default card in a default slot with both ports configured. If I have two cards for storage, I move one port (like e5b to e4b or whatever). After the node is in a cluster (by itself or with other nodes) the storage port modify works just fine. No need to get crazy

modern pendant Dec 19, 2024, 4:23 PM

#

yeah, what should I say? I have had systems where the ports were not configured correctly and we had to use that command. Dunno why. Maybe someone at NetApp forgot to configure it? Or it was quoted/ordered incorrectly? these things happen 🙂

frank pasture Dec 19, 2024, 4:25 PM

#

Any x400 platform with nvme drives will always have a card in slot 5. Both ports are storage ports. Manufacturing could not configure unless it’s this way. More likely the order was munged. I’ve done plenty of installs and never once had the issue with nvme dives

#

This post is about a c800 has internal drives. The card can certainly be modified after the node is in a cluster

#Wait a few minutes, then try the command