importing physical machine into vcenter HA cluster on Vsphere 8.0

[Somnathnandy]

Hi All,

I am a just a biginner of VMware, I am trying to setup a vCenter HA Cluster on vSphere 8.0.2. I am having 3 x Dell R6525 Server and 1 x EMC ME5012 iSCSI Storage with direct connectivity through 10G DAC cable. I have installed configured Vmware ESXi 8.0.2 mapped the volumes into these 3 servers and configured the management networks also. I have installled vCenter Server on one of the 3 Vmware servers and configured it with SSO login with ID administrator@vsphere.local. Now vCenter Management console is opening - I have created DC and tried to create HA cluster with 3 hosts. But found the wizard is creating HA cluster with 2 hosts only, 3rd host just after import getting Disconnected and removed even from inventory. When I am trying to reimport is the same is occuring again. When I checked the monitor founf error message "can not import the host as vm is running on it" but the vm is the vCenter on which I I am trying to create the HA cluster so I am unable to shutdown the VM. also the vCenter is not visible on the cluster created with remaining 2 hosts. Other than this there is no error message on the cluster or hosts inside the cluster. Please help me as it is very urgent to create the HA cluster.

Please help. 

[Somnathnandy]

Hi @Alfista_PS @Marco_Luvisi  @canberkki @a_p_ @TeknoAD 

Thank you for your continiout support to me. I have resolved the problem. First let me tell you that I was trying to implement vsphere HA between 3 hosts with a directly connected shared iSCSI storage based on 10g. All have been purchased from DELL including VMware, which I have told you before. To avoid unwanted driver related and other problems I specifically downloaded DELL customised ISO from VMware site (which I did in my previous installations with HPE hardware and completed without any problems) and this created the actual problem. This time I specifically downloaded ISO of ESXi and VCSA from vmware site (still trial as Dell is unable to provide the actual license because of transition of VMware with Broadcom) and it worked perfectly as you advised before without any problem. Volumes have been mapped properly. DC and Cluster have been created without any problem and all hosts have been imported properly including vCenter and 2 other VMs. But I have other problem which is new to me and shall post now in a new thread. Please help me on this.

Again THANK YOU ALL FOR YOUR KIND SUPPORT.

Regards

Somnath Nandy

[Somnathnandy]

Hi @Alfista_PS 

With new ISO from VMware(previously I was using DELL customised ISO to avoid the driver related problems downloaded from VMware site) all are perfect as you said.

Thank you for your kind support.

Regards

Somnath Nandy

[Somnathnandy]

Hi @Alfista_PS 

Picture attached.

[Alfista_PS · Accepted solution]

Hi @Somnathnandy ,

its great that you have now running and it was pleasure to help you.

By the ay, its a first time that I heard the the vendor customised image done this problems but its good to know 👍.

Please give kudos or mark the helping answer as resolved.

[Somnathnandy]

Hi @Alfista_PS 

As I said before, I have created cluster, imported VMs also after changing the ISO. Now I am facing a networking related problem and I think your help is required here -

I have installed ESXi 8.0.U2 on 3 hosts each having 2 x 32 Core AMD EPYC CPU and 256GB RAM with 2 x 2Port 10g NIC from Broadcom and iSCSI SAN. One 2-port 10g NIC I have used to connect the iSCSI storage(directly connected) for each host, configured vsphere HA and created 3 VMs AD, APP & DB on HA cluster. I want to use the remaining 2-port 10G NIC for these 3 VMs' network connectvity (one vNIC for each VM) for users to connect to these VMs. On each host I have configured one standard vswitch with this 2-port NIC and inside the switch I have configured teaming (algorithm based on IP hash), I have created a port group (10G_DATA) with this standard switch also. There is alternate 1g Network also which is the management network for all hosts and available on the cluster also. When I am creating any VM, in network part this 10G_DATA is available by default but when I am attaching this 10G_DATA with the VM, I am unable to communicate with the outside world, but I am able to communicate with the other VM connedted with this 10G_DATA standard switch. But when I am changing the network to management network from 10G_DATA, I am able to communicate the outside world. the 10G ports are properly terminated on 2 x Cisco 9300 switches(stacked for switch level HA). When I am removing these VMs from 10G network and attaching to 1G network, outside network is able to communicate but there is an continious alarm - "2 load balancing runs skiped" with red tag. when I am again attaching the VMs back it becomes normal. I want to use this 10G network for Data and vMotion.

I am unable to understand how to use this 2-port NIC for my data and vmotion.

Need your help.

Regards

Somnath Nandy

[Somnathnandy]

Hi @Alfista_PS 

Now it has become part 1 and part 2. Part 1 resolved but with part 2 I am confused. Please help. I have already posted in reply.

Regards

Somnath Nandy

[Alfista_PS]

Hi @Somnathnandy,

I know that I ask for it but is the teaming working between hosts and switch and are physical adapters also correctly set and all added? (double check it and try to ping devices)

If yes, have you all set on the vmkernel adapter including gateway and dns? (you can check it on hosts in configuration - networking - virtual switches or vmkernel adapters). Maybe you can also to check in TCP/IP configuration if you have there all needed routings. Maybe you can also check the used VLANs and subnets 🙂

[Somnathnandy]

Hi @Alfista_PS 

I have given the config details of 10G part. Initially I created the vmkernel adapter for this 10G thinking that I can control the services like vmotion, NFC, etc but it vanished from the cluster network. So i have removed the VMK NIC and 10G again appeared in the cluster network.

I am confused fully by this.

Regards

Somnath Nandy

[Somnathnandy]

Hi @Alfista_PS 

In server segment it is only VLAN 0. When VMs are on 1G it is able to communicate with outside. Please find enclosed herewith the logical diagram available on ESXi. 1G & 10G both cases NIC Teaming policy is "Route based on IP Hash" but with 1G, terminated on 2 x Cisco 9300 stacked, I am able to go to internet and with 10G, terminated on same switches, and same NIC teaming policy,  I am not able to go out - confusing me highly. Is my procedure right or wrong for 10G NIC? If the steps taken are right then I shall contact the network vendor.

Regards

Somnath Nandy

[Alfista_PS]

Hi @Somnathnandy,

I dont know what you have done there and how, but I thing I see there a problem.

You have the 10G ports set as they were on other switch (vSwitch) and all should be done on one. And it's good when you do all settings from vCenter, that on all hosts are the same settings. It should look like on picture with 1G ports but you have there also another group with 10G output ports and the VM  Networks group should be directed to 10G output ports and the management to the 1G output ports.

[Somnathnandy]

Hi @Alfista_PS 

10G Network is working w/o any problem - it was kept blocked by network vendor intentionally. After several arguments they have configured it on their switches and it is working. Only part left is Live vMotion which is not happening. By the way, now vmware is licensed including vCenter. Facing a strange problem - after failover, vms are loosing network connectivity. If I change the host(after power down currently as live vMotion is not working) network connectivity is coming back - have you faced this problem? If so please advice.

Thank you in advance.

Regards

Somnath Nandy

[Alfista_PS]

Hi @Somnathnandy,

I thing vMotion will be easier to resolve. You need in VMK (VMkernel adapters) activate needed services which you will use. Or better is to have for some services like vMotion, vSAN, etc to have separate VMK what help you then better work and control over it.

The second problem - I never had such issue. Maybe its related with not correct setting of the vMotion but I thing is still ended if you have correctly configured the networking. My suggestion will be to think what all you will and need to use and how and then look in the network configuration in vmware if all is configured properly. Also will suggest to look on the switches and router for logs if there aren't any logs related the vMotion (use separate ports) which can direct you where can be the problem.

[Somnathnandy]

Hi @Alfista_PS 

I have configured as per the vmware guideline, checked with vmkping and found ok. All the hosts are synced with NTP so there should not be any time difference. Tried 'nc' but sometimes it is working and sometimes it is not working. Unable to understand what is happening. 

Regards

Somnath Nandy

[Alfista_PS]

Hi @Somnathnandy,

ping is OK but it doesnt tell all. Check the logs on the switchers around the time after reboot the hosts when it loose working.

Then check the VMK's if all is set correctly and as I wrote for vMotion and other services you need to enable it for in VMK's to work.

For me it looks that you have some data loops in data flow and the switchers stop the communication after they find it - therefor it work a while after rebooting the hosts.

But from this point I cant help you more, while its needed go deeper and maybe you can post this second problem the Networking group here and there will be people which had an issue like yours and resolved it.