Win2K8 R2 Failover Cluster Private/Heartbeat Network not working

[BostonTechGuy]

Hello,

I have a two node Win2K8 R2 Cluster that I am trying to build.  I cant get the Private/Heartbeat network to work.  Neither nodes will see each across the Private/Heartbeat nic.  This has me totally stumped.  Will layout the information as much as I can:

• Multiple VM Hosts Cluster - ESXi 4.1.0 Update 2
• Hosts using 10G Ethernet to Cisco Nexus 5K Switches connecting back to Catalyst 6K core.
• Custom VLAN on vDS for Private/Heartbeat Nics on both Virtual Nodes
• Network Team states the same VLAN is created on the "Physcial" Network at the core level.
• Private/Heartbeat IP Address for Node One: 10.255.255.1
• Private/Heartbeat IP Address for Node Two: 10.255.255.2
• Subnet Mask: 255.255.255.0
• Regular Public Network on both nodes works fine.
• Can ping fine across Public Network but not Private/Heartbeat, even when both are on same host.

This is all pretty standard as far as I am know?  What did I miss? Completely in the dark why this isnt working.  Have built hundres (really) of physical clusters.  This is first virtual cluster.

Thanks,

Boston TechGuy

[VirtuallyMikeB]

Hey Boston,

Please bear with me, but these things are sometimes small issues.  Something as simple as the hearbeat VLAN not being defined on the Nexus switches would cause these symptoms.  One could define all the VLANs they wanted on the 6000 series core switch, but if they're not defined on the Nexus switches themselves, you're not getting anywhere on those VLANs.

A one-liner as simple as

vlan zzz

where zzz is your MSCS heartbeat VLAN, would suffice on each Nexus switch if this is the problem.  I can see this happening because such a VLAN is not necessarilly a "production" or "user" VLAN, it could be missed during initial configurations.

Also, one has to ensure that VLAN is is allowed on the trunked connections to/from the ESXi hosts on the 10 Gb links.  Something like,

switchport trunk allowed vlan xxx, yyy, zzz

where zzz is, again, your MSCS heartbeat VLAN and, perhaps, xxx and yyy are production or server VLANs.

Cheers,

Mike

http://VirtuallyMikeBrown.com

https://twitter.com/#!/VirtuallyMikeB

http://LinkedIn.com/in/michaelbbrown

Note: I should add, for completeness, that one should always comment in, label, or describe your switch configurations.  To that end, a proper VLAN addition would look something like this:

vlan zzz

   name MY_VLAN

[BostonTechGuy]

Mike & everyone,

Thanks for the info.  It was a number of issues.

You were 100% correct.  The Nexus had local configs that didnt match each other in different segments (phyiscal locations in Datacenter) on the physical network.  We matched all the Nexus to each other and matched them to the core.

Also our AD team stated over and over again that the GPO turns off the local firewall.  Came to find out after ripping the servers apart, the GPO was a hold over from Windows 2003.  So on a Win2008 R2 server that GPO only disables the DOMAIN LEVEL Firewall.  Checked Private and Public, both still enabled.  I disabled them and the HB NICs started talking right away.

Followed up with the AD Team and they were aware of the issue but it has not been a priority to circle back on it..... hours of my life I wont get back. :smileysilly:

Just goes to show you.  It can be anything with Virtual MS Clusters!

Thanks,

BostonTech Guy.

[immanishmittal]

Hi

i have attached some screen shots of  my scenarios. i can ping my host machine ips but cannot pining the heartbeat  IPs 10.10.10.x/8

One vm on this Esxi host 192.168.0.65 and second vm on this Esxi host ip 192.168.0.66. These ips are pining each other.

For or more information please see the attached files.

Thanks

Manish