BladeSystem - General
cancel
Showing results for 
Search instead for 
Did you mean: 

Help with a heartbeat between C7000 chassis

SOLVED
Go to solution

Help with a heartbeat between C7000 chassis

I have a pair of C7000 chassis configured in a fault tolerant cluster under windows 2k8 R2.

We have a pair of VC 10GB fiber modules in each, and a pair of brocade FC modules connected to an EVA 4400.

The VC 10GB modules are connected to separate blades on a 5412.

The two 10GB modules have the same vlan's passed through to them, and the blades are configured to treat them as an active/passive failover team...

What are my options for having a non-switch connected heartbeat between the two chassis?

At this point, I have 4 open interconnect bay's on either chassis.

Thoughts?
14 REPLIES
Johan Guldmyr
Honored Contributor

Re: Help with a heartbeat between C7000 chassis

Hello,

I believe it is possible to connect the OA modules - is this what you are looking for?

Last time I checked they had to be in the same rack for this. This doesn't seem like a hearbeat to me.. but with it you were supposed to be able to manage both chassis in the same place.

What do you want to do with the heartbeat? Is it between two servers or something like that?
The Brit
Honored Contributor
Solution

Re: Help with a heartbeat between C7000 chassis

As far as I know, VC modules in different enclosures can be connected using 10Gb CX4 stacking links, however, this would create a (New) single VC Domain. The new domain would be then be aware of all of the blades and IC Modules in both enclosures.

If this is done, then heartbeat communication between blades in different chassis can take place across the stacking links, just as if the blades were in the same enclosure and using the backplane.

Be cautious however! Initially,the VC Modules in each enclosure form an enclosure VC Domain, so, in your case, you will be starting with TWO distinct VC domains.

I dont think it is possible to "merge" two existing VC domains into a single VC domain. The procedure (as I understand it), involves selecting one of the domains as the "basis" for the new combined VC domain, and then importing the second enclosure. The VC setup from the second (imported) enclosure, i.e. Networks, SANs, Profiles, etc., then have to be recreated in the new combined domain. Or the existing Server Profiles (from the imported domain) have to be modified to use the Networks, SANs etc. predefined in the primary.

I have not been able to find any real clear documentation on this process, and as I said, this is only my best understanding of the process. There maybe members on this forum who will modify my comments, but as I said before, BE CAUTIOUS.

HTH

Dave.

Re: Help with a heartbeat between C7000 chassis

The end goal is to have better inter-chassis communication for a Windows 2K8 R2 cluster (that's running Hyper-V)...

After looking through the VC cookbook (after posting my question here), I noticed the examples for "3.2" which though is designed for ESX, is similar to what I THINK we may need to do.

I took another look at the modules, and determined that we actually have spare gbic's (which is important since I don't have a long enough CX4 cable, but do have spare fiber that could be used for a chassis inter-connect...)

I wasn't part of the chassis configuration... (I'm "just" the network guy), however, I'm trying to gather enough information to make sure this gets resolved in a meaningful manner...

It sounds as though the best plan is to actually merge the VC domains (well, delete the one, and then add it to the other...) it appears that in order to do that, you actually have to stack the OA's as well... (is this correct?).

I know we're passing vlan trunking through, though there's an interesting server configuration defined...

How much pain would take place if I were to delete the one VC domain, join it to the other, and then re-create the server profiles for the blades in the second chassis?
The Brit
Honored Contributor

Re: Help with a heartbeat between C7000 chassis

This really depends on how complicated the configuration is. You would have to recreate all of the networks, Uplinksets, SANs, and profiles from the enclosure being added, in the "primary" domain.

Most of this could be done ahead of time, however you will have to take care with the profiles. Especially if you are using Virtual WWID's an MAC addresses (which are the normal VC configuration choices.) When you create the new profiles in the Primary domain, VC will assign new WWIDs to be use by the HBA's on the new servers. This info obviously has to get to your storage manager so that he can modify the LUN presentations.
MACs are less of a problem. The UplinkSets (if you use them) might need to have the ports reassigned.

Just a couple of things to look into!!

Dave.
The Brit
Honored Contributor

Re: Help with a heartbeat between C7000 chassis

As far as whether or not the OA's HAVE to be linked when using a multiple enclosure VC Domain, I really dont know.

However, just from a routine management perspective, the OA's can be linked to provide management from a single browser window.

If you check between the OA's you will see two RJ45 connectors. In the OA, click on "Enclosure Information" and then open the "Information" Tab for more info.

Dave.

Re: Help with a heartbeat between C7000 chassis

All of our SAN connectivity is currently handled via fiber-channel... I'm not sure that it's impacted by changes to this.

Looking things over again in the cookbook example scenario's...

The Scenario I THINK I want to create might actually be 3:1...

Very similar.

I'm pretty sure I can script a fair amount of this - both for the switch and for the VC.

Am I right in assuming the servers would need to be off-line when I do this? or could I do this with hosts being active? (not expecting connectivity... just wondering aloud).

(This is what I THINK I'd need to do to properly build the VC Domain for these two enclosures...)

Links would be connected on the X1 ports, I'd be teaming dual 10GB up-links at either end of this system... and I'd expect to migrate all servers from enclosure 0 to enclosure 1... and would stage the configuration for the servers in enclosure 1, then alter it for enclosure 2... not sure if I can build the network for "B" without actually defining the up-links until the second enclosure is configured? (This would require some staging to actually implement as there are live VM's between these boxes...)


import enclosure 1.2.3.4 UserName=Adminsitrator Password=blah DoubleDense=False

Add uplinkset VLAN-Trunk-A
add uplinkport enc0:0:X2 Uplinkset=VLAN-Trunk-A speed=auto
add uplinkport enc0:1:X2 Uplinkset=VLAN-Trunk-A speed=auto

Add uplinkset VLAN-Trunk-B
add uplinkport enc1:0:X2 Uplinkset=VLAN-Trunk-B speed=auto
add uplinkport enc1:1:X2 Uplinkset=VLAN-Trunk-B speed=auto

set network network_A SmartLink=Enabled
add network network_A uplinkset=VLAN-Trunk-A

set network network_B SmartLink=Enabled
add network network_B uplinkset=VLAN-Trunk-B

add profile server_0_1 -nodefaultconnection
add enet-connection server_0_1 pxe=disabled Network=network_A
add enet-connection server_0_1 pxe=disabled Network=network_B
assign profile server_0_1 enc0:1

add profile server_0_2 -nodefaultconnection
add enet-connection server_0_2 pxe=disabled Network=network_A
add enet-connection server_0_2 pxe=disabled Network=network_B
assign profile server_0_2 enc0:2

add profile server_0_3 -nodefaultconnection
add enet-connection server_0_3 pxe=disabled Network=network_A
add enet-connection server_0_3 pxe=disabled Network=network_B
assign profile server_0_3 enc0:3

add profile server_0_9 -nodefaultconnection
add enet-connection server_0_9 pxe=disabled Network=network_A
add enet-connection server_0_9 pxe=disabled Network=network_B
assign profile server_0_9 enc0:9

add profile server_0_10 -nodefaultconnection
add enet-connection server_0_10 pxe=disabled Network=network_A
add enet-connection server_0_10 pxe=disabled Network=network_B
assign profile server_0_10 enc0:10

add profile server_0_11 -nodefaultconnection
add enet-connection server_0_11 pxe=disabled Network=network_A
add enet-connection server_0_11 pxe=disabled Network=network_B
assign profile server_0_11 enc0:11


add profile server_1_1 -nodefaultconnection
add enet-connection server_1_1 pxe=disabled Network=network_A
add enet-connection server_1_1 pxe=disabled Network=network_B
assign profile server_1_1 enc1:1

add profile server_1_2 -nodefaultconnection
add enet-connection server_1_2 pxe=disabled Network=network_A
add enet-connection server_1_2 pxe=disabled Network=network_B
assign profile server_1_2 enc1:2

add profile server_1_3 -nodefaultconnection
add enet-connection server_1_3 pxe=disabled Network=network_A
add enet-connection server_1_3 pxe=disabled Network=network_B
assign profile server_1_3 enc1:3

add profile server_1_9 -nodefaultconnection
add enet-connection server_1_9 pxe=disabled Network=network_A
add enet-connection server_1_9 pxe=disabled Network=network_B
assign profile server_1_9 enc1:9

add profile server_1_10 -nodefaultconnection
add enet-connection server_1_10 pxe=disabled Network=network_A
add enet-connection server_1_10 pxe=disabled Network=network_B
assign profile server_1_10 enc1:10

add profile server_1_11 -nodefaultconnection
add enet-connection server_1_11 pxe=disabled Network=network_A
add enet-connection server_1_11 pxe=disabled Network=network_B
assign profile server_1_11 enc1:11

The Brit
Honored Contributor

Re: Help with a heartbeat between C7000 chassis

Regarding FC. It depends on whether you are using VC assigned "Virtual" WWIDs, or "Real" WWIDs in your Server Profiles. If you are using VC-assigned, then they are assigned from a set of WWIDs which are unique to that domain.

When you recreate the Server Profile in the new "Primary" VC Domain, this is a different Domain, so VC will assign different WWIDs. If this happens, then your servers will no longer see their storage.

Your storage manager will have to change the WWIDs that he is presenting the storage to, to reflect the new VC WWIDs assigned to the profile. (This is similar to what used to happen in the old days when an HBA failed and was replaced, all the storage had to be re-pointed to the new WWIDs)

HTH

Dave.

Re: Help with a heartbeat between C7000 chassis

As far as I know, the Brocade switches are configured in one zone, and appear to have a different configuration interface than the VC does.

I'm not sure... and dont' seem to be getting responses from the HP SA's who would be on our account... :(
The Brit
Honored Contributor

Re: Help with a heartbeat between C7000 chassis

Apologies, I assumed that you were using FC VC modules. I have no experience with the Brocades.

Dave

Re: Help with a heartbeat between C7000 chassis

Hmmm... so I'm confused... these plug into the VC modules, but appear to have their own interface (web and ssh...) but from what I can see, the only entries I see at all are related to NS...

(I'm very unfamiliar with this particular device...)
The Brit
Honored Contributor

Re: Help with a heartbeat between C7000 chassis

Hi Scott,
I was wondering if you are having any issues with this configuration?? I jsut noticed,

Add uplinkset VLAN-Trunk-A
add uplinkport enc0:0:X2 Uplinkset=VLAN-Trunk-A speed=auto
add uplinkport enc0:1:X2 Uplinkset=VLAN-Trunk-A speed=auto

Add uplinkset VLAN-Trunk-B
add uplinkport enc1:0:X2 Uplinkset=VLAN-Trunk-B speed=auto
add uplinkport enc1:1:X2 Uplinkset=VLAN-Trunk-B speed=auto

this seems to indicate that you are adding port

:: and :: to Uplinkset VLAN-Trunk-A

Question 1 is what is Bay 0???, Bays are numbered 1 - 8.
Second question is more of an observation. As a general rule, you should not add ports from different VC modules to the same Uplinkset.

Just an observation.

Dave.

Re: Help with a heartbeat between C7000 chassis

That's part of why I was asking... I haven't actually tried that config... right now those links are live, but are an active/passive standby pair... I thought I used the right bay's, but perhaps it's supposed to be 1 and 2...

Thanks!

I've been told that I really need to wait for someone else's approval at this point.

Thanks for your help though! :)
hasrul
Occasional Advisor

Re: Help with a heartbeat between C7000 chassis

ur objective is to have heartbeat connection between 2 c7000 right? for cluster w2k08, u doesn't need the heartbeat connection anymore.without it, ur cluster still can work without any issue

Re: Help with a heartbeat between C7000 chassis

The cluster spans both enclosures, and it works great... until the switch both enclosures are connected to gets restarted for routine maintenance... The hope is to directly connect & stack the virtual connects in both enclosures...

(Also, as a side note, I discovered the hard way that using the "save supportinfo" command while SSH'd into a Flex-10 Virtual Connect prompt will interrupt communication)