• The ability to have tagged and untagged traffic on the same interface started with ONTAP 8.0+
• so, no luck for you with the 2050...
  
• You can *add* vlans non-disruptively to an interface on which you already created at least one vlan. For the initial 'vlan create' though, the interface needs to be down. 
• the 'nfo' parameter just means, that if (one of) these interfaces becomes unavailable, it triggers a takeover. You need to configure 'partner' interfaces (on both nodes) to configure takeover behavior.
• For (almost) non-disruptive reconfiguration you could do the following:
• prepare the /etc/rc on both nodes with the VLAN configs you'll need
• you might want to practice/test with a simulator...
• don't forget to create a failover-interface on the partner (with the same VLAN...) and to configure the switch for that one, too!
• Then either
• failover/giveback both nodes      (then you KNOW it works even in the event of a failure)
  
• or
• do a "source /etc/rc" to 'execute' the rc-file      (you might have to down the interface first, since the commands in the rc assume a non-initialized state of the interfaces...). It might be faster than a takeover/giveback, though.

ONTAP 7.3.6.

This is a FAS2050C (IBM N3600 A20) that has a SAS expansion card in
each controller.  As such, we're limited to only the two onboard GigE
ports.

e0a and e0b both go to different physical switches and suck in two
different networks, so we can't vif 'em together easily.

I have a need to introduce a vlan interface to e0a and am wondering if
it's possible to do so non-disruptively and preserve access to the e0a
via the untagged traffic on the network.  Both interfaces are
configured with 'nfo'.

I've found conflicting information on this.... some say you can mix
untagged on your e0a interface and tagged traffic on e0a-vlan## just
fine.  Others say it's not doable and I'll need to get my network guys
to tag all the traffic and move my e0a to a vlan interface.

Not sure I'm seeing a way to do this without some downtime... even if I
do a cluster failover, I don't see how a failback would work if I've
mucked with the interfaces.

Thanks,
Ray
_______________________________________________
Toasters mailing list
Toasters@teaparty.net
http://www.teaparty.net/mailman/listinfo/toasters