FAS2020 vif and vlan design question

radnorsan · ‎2010-05-25

Hey all

I wanted to ask this question as I am fairly new to netapp. We just bought two fas2020s. They will be used in two seperate buildings as VM storage and CIFS shares. I'm not concerned about CIFS traffic. It will be minimal. VM storage is NFS. What I'd like to do is segment the IP storage (NFS traffic) in it's own vlan, say vlan 101. Obviously, I could use e0a for CIFS and e0b for NFS but I feel like I need the NICs for VMware. So, I'd also like to aggregate the two nics on the filer for both throughut and redundancy. My issue is how to do this with only the two NICs on the 2020 filer.

What I am thinking (not knowing if it will actually work) is to create vif0 with e0a and e0b. Then create two secondary vifs, svif0 and svif1. Then create the vlan with svif0 to get svif0-101. svif1 will stay in the default vlan to allow me to connect and manage the filer.

Any help is appreciated.

rkaramchedu1 · ‎2010-05-25

Yes for the most part.

Create can create a virtual interface (vif) using e0a and e0b. Depending on the switch capabilities, it can be a lacp vif or a multi mode vif. Both types aggregate the links so you have two links to use (2 x 1Gbps links in the vif)
On that vif, you can create a cifs and nfs vlans (if they are indeed in different vlans). Doing so will result in two vlan specific interfaces that you can then configure with IP addresses etc.

radnorsan · ‎2010-05-25

That's what I thought. My HP switch does support LACP. I have been testing creating multi mode vifs but I have been having some difficulty.

That is exactly what I need though. Except, the second controller in the fas2020 is handling the CIFS. Controller A is handling VMware NFS and Controller B is handling CIFS. They are in an active/active configuration. I really just need Controller A to have two multi or LACP vifs with one vif in a NFS vlan and the other vif in the default vlan so I can still manage the filer.

I'm a little unsure how to proceede from here.

If possible, could someone post the proper syntax for the vif and vlan creations?

radnorsan · ‎2010-05-26

Thanks for you help with this. Ok, now I have each controller's 2 NICs configured with a LACP vif and the ports on the switch trunked with LACP. I can ping evrything and I think I'm halfway there.

Next to configure is the vlans. I'll start by saying I'm not a network guy, so this is where I get even more confused. The switch I'm connecting to is an HP Procurve. There are many vlans in our environment. The vlan I am in, and we'll call it our management vlan is vlan 90. I have created a vlan for my VMware (NFS) traffic. This is vlan 101.

Here is a rough layout of how everything currently looks. The two filers are in an active/active config and each vif is the other's partner.

filer a (VMware & NFS Traffic)

e0a e0b

\ /

vif0a

|

____|____

| |

Port 23 Port 24

|________|

|

LACP Trunk on the switch -----> vlan 90 UNTAGGED

-----> vlan 101 TAGGED

-----> All other vlans NO

filer b (CIFS)

e0a e0b

\ /

vif0b

|

____|____

| |

Port 21 Port 22

|________|

|

LACP Trunk -----> vlan 90 UNTAGGED

-----> vlan 101 TAGGED

-----> All other vlans NO

I guess my first question is, do I need to create any vlans in the filer, i.e. "vlan create vif0a 90 101"?

And, I'm not sure what my other questions are yet... thanks for your help in advance.

radnorsan · ‎2010-05-26

I just wanted to give an update on where I'm at with this and the steps I took to get there. This is what it looks like now.

filer a (VMware & NFS Traffic)

e0a e0b

\ /

vif0a

____|____

| |

vif0a-90 vif0a-101 vlans 90 and 101

|________|

|

Port 23 Port 24

|________|

|

LACP Trunk on the switch -----> vlan 90 TAGGED

-----> vlan 101 TAGGED

-----> All other vlans NO

filer b (CIFS)

e0a e0b

\ /

vif0b

____|____

| |

vif0b-90 vif0b-101 vlans 90 and 101 (vlan 101 is there just in case of a takover)

|________|

|

Port 21 Port 22

|________|

|

LACP Trunk -----> vlan 90 TAGGED

-----> vlan 101 TAGGED

-----> All other vlans NO

Commands to get to this point:

From BMC

bring down all nics - ifconfig e0a down / ifconfig e0b down

vif create lacp vif0a -b ip e0a e0b

vlan create vif0a 90 101

ifconfig vif0a-90 10.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

ifconfig vif0a-101 10.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

Repeat for controller b

From BMC

bring down all nics - ifconfig e0a down / ifconfig e0b down

vif create lacp vif0b -b ip e0a e0b

vlan create vif0b 90 101

ifconfig vif0b-90 10.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

ifconfig vif0b-101 10.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

vlans and network configuration seem to be working correctly on the netapp side of things. Now it's on to VMware virtual networking and vlans. Thanks for the help. Do I seem to be missing anything?

rkaramchedu1 · ‎2010-05-26

vif create lacp vif0a -b ip e0a e0b

vlan create vif0a 90 101

ifconfig vif0a-90 10.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

ifconfig vif0a-101 10.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

Looks good overall. I'll let others chime in and give their feedback. I would add couple of notes, however.

if you want the partner filer to take over these interfaces in an event of a CF event, you might want to stick that "partner <name>" option at the end of the ifconfig statement. Such as (ifconfig vif0a-90 10.x.x.x netmask 255.255.255.0 mediatype auto partner vif0b-90)
Also, for the VIF that is designated for NFS, you might also want to turn off WINS, since that interface will only carry NFS traffic. Such as ((ifconfig vif0a-101 10.x.x.x netmask 255.255.255.0 mediatype auto partner vif0b-101 -wins)

HTH

rorzmcgauze · ‎2011-02-11

HI there,

I was wondering what you did from the switch side.

I have been having an issue with a HP Procurve 2810-48G J9022A. I have the trunk set and thats working fine along with LACP. the problem comes when I implement the vlans. I started with just one vif1-100 that I was going to enable jumbo frames on and use for iscsi. then another for what is in essence the native vlan (tho i believe using an id like 9999 will solve this).

I have tried several thing from both the web and cli interface for the switch but no avail and at the moment am running with no vlans. each time the vif would not come up.

The issue is now the system is on an internal vlan for iscsi use and has no way of ASUP. Any help would be great thanks.

radnorsan · ‎2011-02-11

Hi. Let's see if we can help. I'm assuming you created the vlans on the switch first. If not, you need to create the vlans on your 2810. Then trunk your ports you want and tag the trunks for the appropriate vlans. If you tag the ports and then trunk them, it won't work as the vlan tag will dissappear. You need to trunk first, then tag. So let's assume vlan 100 is iscsi and vlan 10 is your default vlan. Trunk your ports and tag them on vlan 100 and vlan 10

Ruairi McBride wrote:

then another for what is in essence the native vlan (tho i believe using an id like 9999 will solve this).

^^ Also, the standard vlan range is 1 - 4095. I don't know what 9999 will do, if it will work at all.

What you can do to get ASUP to work is first create a vif on the netapp side with whatever nics you want.

For example, do a

"vif create lacp vif0a -b ip e0a e0b" >>> this will give you a virtual interface

Then create your two virtual interfaces in the vlans you tagged to the trunk on the switch for your netapp, in this case vlan 100 and 10.

This would be something like the following:

vlan create vif0a 10 100

ifconfig vif0a-10 x.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

ifconfig vif0a-100 x.x.x.x netmask 255.255.255.0 (fill in actual IP address and correct netmask)

Now you should have a vif named vif0a that is comprised of vif0a-10 and vif0a-100. Hope this helps. Let me know if you have any other questions.

rorzmcgauze · ‎2011-02-16

That is what I am trying to do and from the FAS2020 point of view I have that all sorted and has worked fine for me in the past. The issue is the HP switch for some reason, the trunk works fine but when I put VLANs into the mix on the switch it stops working. My issue is that you have to use the CLI menu system that I am unfamiliar with.

radnorsan · ‎2011-02-17

Perhaps your switch is bad? I can tell you though that you don't need to use CLI on the Procurve to do this. The menu will allow you to do everything necessary like creating the vlan and tagging the appropriate ports. Once the vlan is created you need to tag the correct ports. Check out the picture. Rnlan is the vlan I am in and vmnfs is the vlan I created for VMware NFS traffic. My FAS2020 has two controllers. Trunk 1 is the trunk the first Netapp controller plugs into and Trunk 2 is for the second controller. You need to tag both vlans for the two trunks because the trunk has virtual interfaces in both vlans. Next, Trunk 4 and 8 are my physical NICs in my ESXi hosts. These only need to be on the vmnfs vlan, so I set this vlan as untagged for these trunks so all traffic is on this vlan.

As a side note, LACP trunking is not compatible with ESX. You need to configure it as multi. Hope this helps.

p_maniawski · ‎2011-02-22

Remember with ONTAP 7 series you can't mix tagged and untagged VLAN in one vif (it's avaiable since ONTAP 8). Both must be tagged. This is the reason we had to create artificial VLAN (tagged) and route it to default VLAN (LAN), which can't be tagged using our switch (Allied Telesis drawback).

ademuynck · ‎2011-07-05

Hi,

With Allied Telesis switch, you have to upgrade your switch firmware, and then create a new default vlan.

This method let you tagged your default vlan.

So, yes you have to create VLAN on your filer, and tagged port on switch corresponding to your filer.