"Failed to add witness host to a stretched cluster" when vSAN separated from Management Network.

[3apa3a_b_ta3e]

I experimented with latest vSphere 6.5 vSAN and have some problem.

When i tried to build my first test Stretched Cluster everything was cool - i just setup two data node, bind vSAN to vmk0 (same as Management Network), deploy Witness Appliance and connect both vmkX to the same network - same VLAN, same switch. And that works.

After that i choose to reconfigure whole vSAN - reset config of both ESXi, deploy new Witness on separate host, create separated VLAN, vmkX on nodes and additional VM Network on host, where Witness resides, apply different ip addresses for vSAN network. Checked connection with vmkping from vSAN-vmk to new ip-address - connectivity is ok. But i get error "Failed to add witness host to a stretched cluster" when try to configure ROBO cluster from WebGui. Whitness is configured with right "Sub-Cluster UUID" but cannot see other nodes - "Sub-Cluster Member Count: 1".

Ok, then i updated Witness Appliance with Update Manager to the latest version (other hosts already have that build), added unicastagent address to both data-nodes via cli (there was NO unicastagent before) and try to configure Stretched Cluster again via Gui. Still no luck. Whitness shows as not configured - "vSAN is not enabled".

Then i try to add Witness via cli - hmmm, success, "Sub-Cluster Member Count: 3" and "Local Node State: AGENT". But WebGui still shows "Stretched Cluster status Disabled".

AFAIK, manual said - "VSAN VMkernel traffic should be isolated to a layer 2 non-routable VLAN", right? I put vmk1 of Witness into this isolated VLAN.

What i do wrong?

[3apa3a_b_ta3e]

Anyone?

[TheBobkin]

Hello,

Regarding issues in GUI:

Are all hosts and vC on the same build-level?

When you are adding Witness host to cluster from the CLI you have to specify that node-type as Witness with '-t' (or --witness-node) and specify a Preferred FD:

#localcli vsan cluster join -u <UUIDOfCLuster> -t -p <PreferredFaultDomain>

Bob

-o- If you found this comment useful please click the 'Helpful' button and/or select as 'Answer' if you consider it so, please ask follow-up questions if you have any -o-

[3apa3a_b_ta3e]

All ESXi (include witness appliance) - build 5310538.

vCenter - 5178943.

Yes, i know about it, my string was

esxcli vsan cluster join -u 52500415-ee32-e90c-e273-d20228f9f791 -t -p Primary

Still no luck.

[TheBobkin · Accepted solution]

Hello,

Okay, so you may have a problem there, you are using vCenter of lower version than the hosts:

The hosts are using vSAN 6.6 - ESXi 6.5.0d (Build:5310538)

But vCenter is using 6.5 - vCenter Server 6.5 0b Patch 1 (Build:5178943)

Update the vCenter to vCenter Server 6.5 0d Express Patch 2 (Build:5318154) and then try to configure this (from GUI first and CLI if that fails).

Bob

-o- If you found this comment useful please click the 'Helpful' button and/or select as 'Answer' if you consider it so, please ask follow-up questions if you have any -o-

[3apa3a_b_ta3e]

Bingo!

Everything is worked after upgrade vcenter!