Thanks Tim/Francis. Yes we are using interface groups, I was just trying to simplify things a lot.

I'm still confused though because that is how we currently have it set up, yet the VIF still fails to migrate. In the messages log, it states "ifconfig: 'vif_orig_data' cannot be configured: Address does not match any partner interface."

Here is filer1's (active) /etc/rc:
hostname filer1
ifgrp create single vif_orig_data e0b e0a
ifconfig e0M `hostname`-e0M netmask 255.255.0.0 mtusize 1500
ifconfig vif_orig_data `hostname`-vif_orig_data netmask 255.255.255.0 partner vif_orig_data mtusize 1500 trusted -wins up
route add default 10.10.10.254 1
routed on
options dns.enable on
options nis.enable off
savecore


And here is filer2's (standby) /etc/rc:
hostname filer2
ifgrp create single vif_orig_data e0b e0a
ifconfig e0M `hostname`-e0M netmask 255.255.255.0 mtusize 1500
route add default 10.10.10.254 1
routed on
options dns.enable on
options nis.enable off
savecore

When I read this line from the ifconfig man page "A network interface performs this role if it does not have a local IP address but a partner IP address, which you assign by the partner option of the ifconfig command" it sounds to me like you would run "ifconfig vif_orig_data partner vif_orig_data" on the standby node, and remove the partner option from the active node.


On Tue, Oct 6, 2015 at 5:14 PM, Francis Kim <fkim@berkcom.com> wrote:
Phil,

You’re correct.  If filer2 has no data volumes, then there’s no good reason to have a “data IP” on filer2.

The ifconfig statement you really want in your /etc/rc is something like the following:

filer1> ifconfig e0a <IP-address> netmask <mask> partner e0a

This lets filer1’s IP address move to filer2’s interface e0a during a takeover.  While using the same interface (e.g. e0a) in both controllers is not required, this symmetrical use of interfaces is good management practice.

Your suggestion of “ifconfig e0a 10.10.10.10” on filer1 means “Let interface e0a of filer1 takeover the IP address 10.10.10.10 of filer2 for the duration of the takeover”, which doesn’t sound like what you want.

Since you did mention VIF, while your example only shows bare ports, you should remember to create the ifgrp/vif on both filer1 and filer2 if you’re going to use that.

More comprehensive explanation is here.  https://library.netapp.com/ecmdocs/ECMP1210206/html/GUID-525FCF4E-109D-4E6F-9256-C6E56438CA9A.html
Francis Kim

On Oct 6, 2015, at 1:54 PM, Philip Gardner, Jr. <phil.gardnerjr@gmail.com> wrote:

Hi all -

I have a bit of a throw-back question. I swear I've read all the docs availalbe (HA 7-mode config guide, system admin guide, network mangement guide, ethernet storage design TR) but I am still confused on how to properly configure a 7mode HA VIF failover in a situation where you only have one VIF that needs to float to a standby controller. Support tells me this is "not a common configuration" and that the solution is to have a data IP on both controllers, even if both controllers are not being used in active/active. I realize that I can do an ifconfig on each controller, with the partner option specifying the other node's IP/interface. However, this particular filer is just a large aggregate, single volume, single NFS access IP. So I don't see a point to having a data IP on the other controller...

According to the ifconfig man page (https://library.netapp.com/ecmdocs/ECMP1196979/html/man1/na_ifconfig.1.html), this is possible:


"On a node in an HA pair, a network interface performs one of these roles:
...

A standby network interface for the partner. That is, if the partner fails, the network interface works on behalf of the partner. When the node is not in takeover mode, the network interface is idle. A network interface performs this role if it does not have a local IP address but a partner IP address, which you assign by the partner option of the ifconfig command.

The node maps a partner IP address to a shared or standby interface when the node initiates a takeover operation. In takeover mode, all requests destined for the partner IP address are serviced by the shared or standby interface. Also, in partner mode, if a command takes a network interface name as an argument, enter the network interface name of the failed node. The command is executed on the shared or standby interface on the live node. Similarly, in partner mode, a command for displaying network interface information displays the network interface name of the failed node, even though the command is serviced by the shared or standby interface on the live node."


Help me think this through. This means if I want filer1 to be active and filer2 to be standby with a sinlge VIF, filer1 would have something like 'ifconfig e0a 10.10.10.10', while filer2 would have something like 'ifconfig e0a partner 10.10.10.10'?

--
GPG keyID: 0xFECC890C
Phil Gardner
_______________________________________________
Toasters mailing list
Toasters@teaparty.net
http://www.teaparty.net/mailman/listinfo/toasters




--
GPG keyID: 0xFECC890C
Phil Gardner