Strange network behavior

[OmniTech]

I tried testing HA on a 2 hosts cluster. Everything went as expected beside one minor thing.

Please, read carefully.

I tried a continuing ping on 2 VMs on Host A, on 1 VM on Host B and on Host A itself.

When I shutdown Host B, the Vm it owns was restarted on Host A, losing pings on this VM until it came back online.

I did lost any pings on the other devices.

But when I restarted host B, I did not lost any pings from the SC of Host A, but as ESX was booting on Host B, I lost 3-4 pings from the VMs on Host A (wich was untouched).

How in the world is that possible?

[OmniTech]

Oh, I almost forgot.

On Host A, I get alarms:

"network uplink redundancy lost"

"Network connectivity lost"

No errors on HA logs.

Also, I lost 12 pings on VMs on Host A, not 4-5.

[amvmware]

how are the nic's configured on each host to provide resilience and redundancy to the service console.

[OmniTech]

There is 2 SC with 2 Nics each on 2 seperate vSwitch. The vSwitch is on Failover Team.

The VMs use a diffenrent vSwitch with, obviously, diffenrent nics.

As I mentionned, I do not lose connectivity to the SC of the Host, only to the VMs

[amvmware]

I am wondering if the issue is to do with host B booting and HA is a secondary problem.

I would recheck the config of host B and the network ports it is connected to to make sure they are all configured correctly. This includes, vLAN config, tagged ports...etc

[OmniTech]

I could see if HA is the problem by disabling it, and do the test again.

[amvmware]

I would agree - From what you have described it would appear the VM network on host A is being impacted by the reboot of host B - it might be worth bring down host A and see if the same issues occur on host B.

[OmniTech]

Just did the test by disabling HA first.

No pings were lost.

It seems to be related to HA when the second host reboots... But how, this I don'T get.

[amvmware]

and what happens if you reboot host A

[OmniTech]

Sorry, I did get the same Alarms on the host that is not rebooted.

But I did'nt see if the pings were lost.

[amvmware]

I would review your physical network configuration for both these hosts and make sure the physical network configuration is correct on the network switche(s) - it looks to me like you have a network issue.

[OmniTech]

But why would it affect only VMs and only when HA is enable?

[amvmware]

The only thing i can think is there is a misconfiguration on the network switch either ports are not tagged correctly and they think they are on the correct vLAN - we can guess at the reasons but until you start checking you are never going to know.

[OmniTech]

It was a VLAN 1 problem. VMWare told me that ESx does'nt like VLAN 1