VMware Cloud Community
DefenderAtkins
Enthusiast
Enthusiast
Jump to solution

Distributed Switch kernel IP addressing dilemma

Hi all,

I've got a strange, perhaps a silly question concerning VDSwitch kernels.

I have two VDSwitches which are assigned to one vmnic each: VDSwitch1 > vmnic4 and VDSwitch2 > vmnic5. In each VDSwitch, I have management portgroup: MGMT1 and MGMT2.

I want to add a host to both VDSwitches for management redundancy. When I added it, its host IP 10.20.64.66 appears in one VDSwitch as the kernel in MGMT portgroup.

There is no kernel in the MGMT portgroup of the other VDSwitch so I tried to create the kernel for it. I can give it an IP in the same range as the host IP. Is this correct?

What I'm confused about is because the first kernel is the host IP, does the second kernel also have to be the same IP?(which makes no sense to me though)

Thanks

pastedImage_0.png

0 Kudos
1 Solution

Accepted Solutions
daphnissov
Immortal
Immortal
Jump to solution

I think firstly is the important question: Why do you feel it's necessary to have management vmkernel redundancy at the IP address level? This is something which is really not a good idea and definitely not common. What's far more common, however, is to have management redundancy at the vmnic level (physical NIC) but not two separate IP addresses assigned from the same (presumably) L2 segment.

Secondarily to this (and not related), but the patch rev of your ESXi host suggests it is horrendously outdated within the 5.5 code branch. Regardless of what you decide to do, you really, really should patch that up to the latest available build even if you stay with 5.5.

View solution in original post

0 Kudos
3 Replies
daphnissov
Immortal
Immortal
Jump to solution

I think firstly is the important question: Why do you feel it's necessary to have management vmkernel redundancy at the IP address level? This is something which is really not a good idea and definitely not common. What's far more common, however, is to have management redundancy at the vmnic level (physical NIC) but not two separate IP addresses assigned from the same (presumably) L2 segment.

Secondarily to this (and not related), but the patch rev of your ESXi host suggests it is horrendously outdated within the 5.5 code branch. Regardless of what you decide to do, you really, really should patch that up to the latest available build even if you stay with 5.5.

0 Kudos
DefenderAtkins
Enthusiast
Enthusiast
Jump to solution

That makes total sense. I don't need two VDSwitches. 

Now I have one VDSwitch with Portgroups and their kernels, and 3 x physical NICs for redundancy (vmnic0 ethernet, 4 and 5 fibre)

Earlier, the Hosts had a warning sign when HA is switched on because vSphere regarded vmnic 4 and 5 as one device which is true; it is one card with two ports.

So I patched the ethernet port just for management vlan and added it into the physical NICs and it's all working fine.

0 Kudos
DefenderAtkins
Enthusiast
Enthusiast
Jump to solution

And yes, I will also update it. Right now, the management VLAN doesnt have internet access. I'll fix that and do the updates.

0 Kudos