VMware Cloud Community
andreasjan1
Contributor
Contributor
Jump to solution

FT with 2 nodes and 1 SAN

Hi, I have read some threads in this forum, but could not find answers to my questions. I'd like to configure a single FT VM on 2 nodes and 1 SAN.

Question 1, if node 1 (happens to be the primary node with the VM running) is down for few minutes or hours, node 2 will take over the FT VM, but it cannot make the VM redundant. What will happen if node 1 comes back online (without any changes) to the cluster, is this allowed? if node 1 is allowed to join the cluster back, will node 2 start replicating the FT VM to node 1? to the same VM or a new VM?

Question 2 is the opposite of case 1, node 2 (secondary node, no VM running) fails, no failover occurs. If node 2 is allowed to join the cluster back, will node 1 start replicating the FT VM to node 2? to the same VM or a new VM?

Question 3: What will trigger failover, will it occur only when there is complete hardware failure? or will the failover be triggered by simple events such as:

  • one local disk failure, even though RAID still kicks in, i.e. storage is still online
  • RAM issue
  • any one network interface is down
  • one fan failure, etc.

Thanks.

BR, Andreas

1 Solution

Accepted Solutions
scott28tt
VMware Employee
VMware Employee
Jump to solution

Back in my days as an instructor, we used to setup 2-node FT in every vSphere Install Configure Manage class. I've done it on vSphere 4 and vSphere 5 - it works just fine.

If the host running your primary VM fails, the secondary VM becomes your primary - at that point your workload is no longer protected. When the failed host recovers, a new secondary VM will get created on that host, and your VM is protected once more.

Back to the start.

If the host running your secondary VM fails, your workload is no longer protected. When the failed host recovers, a new secondary VM will get created on that host, and your VM is protected once more.

FT doesn't care which host was originally running the primary VM and which was running the secondary VM - it just tries to make sure you have both VMs running so that your workload is protected.


-------------------------------------------------------------------------------------------------------------------------------------------------------------

Although I am a VMware employee I contribute to VMware Communities voluntarily (ie. not in any official capacity)
VMware Training & Certification blog

View solution in original post

7 Replies
lucasbernadsky
Hot Shot
Hot Shot
Jump to solution

Hi, andreasjan1.

In the following PDF you will find the answers you are looking for about failure scenarios in 2-node VSAN deployments. https://storagehub.vmware.com/export-to-pdf/vsan-2-node-guide/7103

Question 1 - HA event will happen. vSAN will replicate the deltas from node 2 node 1 and the VM will be compliant and reduntandt again

Question 2 - Same thing happens to node 2: Once it is alive again, vsan will replicate node 1 data to node 2 and the VM will be compliant again.

Question 3 - Failover will occur when esxi is not able to provide compute power to the VM. (HW, network, OS failure). The exact same reasons as in a traditional infrastructure.

If a disk fails in node 1, the VM will continue to run from node 2 data, if FTT=1 configured, so it won't need to failover.

Hope this answers your questions correctly.

Regards

Reply
0 Kudos
andreasjan1
Contributor
Contributor
Jump to solution

Hi lucasbernadsky,

Thanks for your response. In my case, I am not using vSAN. Secondly, the feature is fault tolerance (FT), not HA.

Reply
0 Kudos
IRIX201110141
Champion
Champion
Jump to solution

OP asks for FaultTolerance rather than vSAN i think.

I have a Dell VRTX with 2 Blades and one pCPU each for exactly that purpose. Customer have one single important VM and a couple of others.

I cant answer the question of the secondary VM comes backup automatically in such a setup. In all my larger setups HA will restart the secondary VM on another Host but in a two Host Cluster this is not possible.

Info: FT 2.0 creates a copy of the vDisks also so think about to present 2 datastores to the cluster and not only one. It will work with one also but i dont like it.

Question 2:

No i cant remember exactly but there is no automatically Failback of the Role. If an HA event occurs which effects the primary VM the secondary just kicks in.

Question 3:

FaultTolerance used HA. So all what is covered by HA.

- For Networkcable you need redundant cards/cable/pSwitch

Regards,
Joerg

Reply
0 Kudos
andreasjan1
Contributor
Contributor
Jump to solution

Hi Joerg,

Yes, I feel the same that it better to create the VMDK (virtual disk) on the local storage of the node rather than on the shared storage (SAN).

In VMware doc, it actually mentions what will happen when the secondary host (node 2) restarts: "This task results in the termination of the Secondary VM that provided Fault Tolerance protection for the selected Primary VM. A new Secondary VM is started, placing the Primary VM back in a Protected state." It seems that it will create a new secondary VM on available node.

I understand that VMware recommends to have 3 nodes in the cluster for FT solution. So for the questions 1 & 2, I wonder if anyone has experience deploying FT VM with only 2 nodes.

BR, Andreas

Reply
0 Kudos
scott28tt
VMware Employee
VMware Employee
Jump to solution

Back in my days as an instructor, we used to setup 2-node FT in every vSphere Install Configure Manage class. I've done it on vSphere 4 and vSphere 5 - it works just fine.

If the host running your primary VM fails, the secondary VM becomes your primary - at that point your workload is no longer protected. When the failed host recovers, a new secondary VM will get created on that host, and your VM is protected once more.

Back to the start.

If the host running your secondary VM fails, your workload is no longer protected. When the failed host recovers, a new secondary VM will get created on that host, and your VM is protected once more.

FT doesn't care which host was originally running the primary VM and which was running the secondary VM - it just tries to make sure you have both VMs running so that your workload is protected.


-------------------------------------------------------------------------------------------------------------------------------------------------------------

Although I am a VMware employee I contribute to VMware Communities voluntarily (ie. not in any official capacity)
VMware Training & Certification blog
IRIX201110141
Champion
Champion
Jump to solution

Yes, I feel the same that it better to create the VMDK (virtual disk) on the local storage of the node rather than on the shared storage (SAN).

Oh.. .i think you miss understood me. The Dell VRTX offer a shared LUN to both nodes and in my case it provided 2 of them.  You cant enable FT for a VM which lives on local storage because FT leverage HA and you need shared storage.

I have a couple of customers which use FT on a 2 node cluster.

Regards,
Joerg

Reply
0 Kudos
andreasjan1
Contributor
Contributor
Jump to solution

Thanks for the clarification.

Reply
0 Kudos