vSpere HA and isolation address for VSAN

[Bab]

According to documents recommendation is change isolation address to vsan vmkernel Gateway and said follow reason :

 

Always set the isolation response to power-off, this would avoid the scenario of duplicate MAC address or IP address on the network when VMs
are restarted when you have a single network being isolated for a specific host.

1- What does duplicate MAC here ? Why will happen duplicate MAC ? Why we have to select power off and restart ?

2- When we change isolation address to vsan vmkernel  it means  ping "source = vsan vmkernel to destination = isolation address (vsan vmkernel gateway)"  ?

3- I am using both type storage in my environment traditional storage (shared storage ) and VSAN now if I change isolation address to vsan vmkernel gateway now if isolation apply on management . what happen will be do because the isolation address is VSAN gateway ? is that means we can just check isolation on ONE vmkernel (management or vsan) ?

4- As I know isolation in vsan means for example vsan network unplugegd on one host and in this mode it will apply according to our role )such as power off and restart vm on other host). now want to know this is like failure on that host is that correct ? also if the host fail vms will be restart on the other hosts

 

 

 

 

[Duncan Epping]

1. A duplicate ip or mac address could happen during an isolation event because when you do not select "power off' as the isolation response, HA may be able to restart the impacted VMs, yet the VMs are not powered off, thus resulting in two instances of the same VM running
2. you do not change the isolation address to the vsan vmkernel, you change the isolation address preferably to the default gateway on the vSAN network
3. when vSAN is enabled the HA traffic will use the vSAN network, even when you also use regular storage. As such you will need to use an IP address on the vSAN network for isolation detection
4. yes, this is a type of failure event

[Bab]

 

A duplicate ip or mac address could happen during an isolation event because when you do not select "power off' as the isolation response, HA may be able to restart the impacted VMs, yet the VMs are not powered off, thus resulting in two instances of the same VM running

 

1- what is your means from second instance here ? is that your means for example FTT=1  "VM on host is isolate but we did not set power of and restart now according to storage policy it will up another on the host ? would you please give an example for this ?

 

you do not change the isolation address to the vsan vmkernel, you change the isolation address preferably to the default gateway on the vSAN network

 

2- ok . Now after change isolation address the ping for check isolation will do from VSAN vmkernel as source and isolation address as destination ?

3- What is difference between a host failure happen in vsan cluster or isolation happen in vsan cluster (while set on power of and restart ) ?

[Duncan Epping]

1)
No, the second instance has nothing to do with FTT=1 really. This can happen also with traditional storage. If you do not have the Isolation Response configured then you can end up in the following situation:

VM runs on Host 1, Host 1 loses access to the vSAN and Management Network, Host 2 sees host 1 is missing and restarts VM, because the isolation response is not configured, the VM will also still be running on Host 1. On Host 1 the VM has no access to storage, but it is still accessible on the VM network.

2) Yes correct

3)
Difference between a host failure and an isolation? The difference is that with isolation HA is going power off the VM as well. With a host failure we don't need to do that.

[Bab]

1- You said duplicate MAC is not related to VSAN but really confused . for example in non VSAN environment vm1........vm10 is on host1 now host1 cannot ping isolated address so it is in isolation status.  now I set isolation response = disable . now what is second instance here ? I set disable isolation thus nothing happen

 

2- You said difference between host failure and isolation in VSAN is just isolation response .

but want to know also in host failure VMS will restart and start on the other hosts in both mode the vm will be restart on the other host Is that correct ?

[Duncan Epping]

 I would suggest you read my book: https://www.rubrik.com/en/resources/white-papers/19/clustering-deep-dive-ebook

What you are forgetting is that when a host is isolated and you set the isolation response to disabled it doesn't mean that HA will not restart the VM. If the host does not respond to anything, than HA does not know if it is isolated or dead. All of this and more is described at length in the above ebook, which you can download for free.

And yes, with an isolation and with a host failure HA will restart the VMs

[Bab]

So thanks

but my means is " When happen host isolation and we set power off and restart vm "  and when happen host failure  there is no difference in this mood . in both  the VM will restart on the other host .

Is that correct ?

 

 

[Duncan Epping]

And yes, with an isolation and with a host failure HA will restart the VMs

[Bab]

One question now imagine in our cluster we have shared storage and VSAN now VM1, VM2, VM3 , VM4 are on VSAN and VM5,VM6,VM7,VM8  are shared storage but all of them reside on same host and change isolation address to vsan gateway vmkernel also I set power off and restart vm for isolation action . now vsan network failed so it detects isolation happened now have to power  off and restart vms want to now it will restart VM1, VM2, VM3 , VM4 or will restart VM1, VM2, VM3 , VM4, VM5,VM6,VM7,VM8 ???

 

 

[Duncan Epping]

all

[Bab]

Don't think this is so bad . because isolation happened just for VSAN network (VM1.....VM4) although it will reboot all vms on that host