VMware Cloud Community
jim33boy
Enthusiast
Enthusiast
Jump to solution

VSAN configuration 2 node

Hello,

I recently set up a 2 node VSAN cluster direct connect using direct connect 10Gb between the hosts.  I may have set something up incorrectly as when I go to fault domains it shows as "stretched cluster". Can you tell me if this is normal in a 2 node direct connect configuration? If not can you please help me with the steps to change from a stretched cluster to a 2node direct connect?

Thank you

Reply
0 Kudos
1 Solution

Accepted Solutions
depping
Leadership
Leadership
Jump to solution

Nothing wrong here, two node + witness from an architectural point of view resembles a stretched cluster, that is why it shows up as stretched. So just leave it as is, nothing wrong with your configuration!

View solution in original post

Reply
0 Kudos
4 Replies
depping
Leadership
Leadership
Jump to solution

Nothing wrong here, two node + witness from an architectural point of view resembles a stretched cluster, that is why it shows up as stretched. So just leave it as is, nothing wrong with your configuration!

Reply
0 Kudos
jim33boy
Enthusiast
Enthusiast
Jump to solution

Thank you very much for the reply and confirmation, I am a big fan of your work.

On a related note, I have 2 directly connected 10Gb NICs(vmnic0 and vmnic1) between the 2 hosts for vmotion and vsan traffic configured in a vds. I went through the VMware VSAN Network Design on storage hub. In regards to teaming and failover, the guide states that if you have any doubts, use a basic NIC team of an active/standby configuration with explicit failover, I am doing this. Vmotion is active on nic0 and standby on nic1 while  vsan is active on nic1 and standby on nic0.

The guide has suggestion based on what appears to be switched scenarios. I an using direct connect and I am seeking clarification on the following.

  • Notify switches: By default this is selected, should this be left at yes in a direct connect configuration?
  • Fail Back: The guide states that in an active/active you are best served setting this to no. Because I am using direct connect with active/standby, is it suggested to set fail back to no? I would like the traffic to fail back to the configured NIC but not at the cost of performance impact.
  • I using direct connect I am passing vsan traffic and vmotion traffic over the same physical NICs. Should I be using VLAN IDs on the vsan and vmotion portgroups? Or is this unnecessary because the vsan vmkernel portgroup is explicitly checked to pass only vsan traffic and the vmotion vmkernel port group is explicitly checked to pass only vmotion traffic?

Thank you

Reply
0 Kudos
depping
Leadership
Leadership
Jump to solution

Good morning,

I wouldn't worry about notify switches, leave it set to default, there's nothing to notify here, but it doesn't hurt either. In terms of Fail Back, I am on the fence. If you have intermittent failures you will see the vmkernel interface flip flopping between those two NICs, I would rather be alerted about the change and investigate then seeing the vmkernel interface move back and forth and hurting performance. In terms of VLAN IDs, there's nothing to configure here, it it a direct connection, physically isolation, so no need for logical networks.

Reply
0 Kudos
jekoy
Contributor
Contributor
Jump to solution

Hi,

If I may ask, what cable are you using for direct connect? 

Are you using crossover cable or just straight cable Cat6?

 

Thanks,

Jek

Reply
0 Kudos