VMware

HP Hosts not showing physical adapters in vcenter

I’ve deployed a new vcenter, photon appliance, 6.5 u1, used migration assistant coming from a physical hp server with vcenter 5.5, 2 hosts are also on 5.5, installed back then with HPE ISO.

After upgrading, the cluster for the two hosts shows no HA available. vcenter cannot see the physical adapter into the VDS, I cannot find the 2 vmkernels either. I can ping them, I can ping the VMs running (still up in their own VLANs), but in new 6.5 venceter, they don’t show. I tried to reboot a VM, it doesn’t even come back up because the host’s state doesn’t allow it.
I can use the cli on ssh console in the host to boot it back again.

The hosts are HP DL385p Gen8, they have multiple cards quad pci and quad onboard, all of them into a distributed switch in the cluster.


View Reddit by ditaccioView Source

Related Articles

3 Comments

  1. Possibly similar, one of my ESXI hosts does not like the mellanox card I have. I have to shutdown the host, unplug power and reconnect. As a result I avoid warm reboots as much as possible, updates be damned.

  2. I resolved this by removing hosts, restart all the agents on the hosts with “services.sh restart”

    also deleted a giant log file in /tmp before upon agents restart, we couldn’t connect directly to the host (503 service unavailable)

    deleted the big log “mili….log” we restarted the services properly again and then we re-added the hosts.

    Rebooting was not a solution as vcenter 6.5 appliance was running on one of them and vmkernel for migration wasn’t available.

    Once the host was added in vcenter, we were able to see adapters, but they didn’t have VDS assigned

    after assigning the vmnic to the dvuplink in the VDS (adding/managing hosts), they needed to re-assigned to the dvuplinks

    once assigned we migrated the vmkernel into the mgmt vlan

    it seems now it found HA heartbeat again, but the VMs didn’t show inside the VDS, we found out “editing” each VM, without touching anything, clicking ok, was putting them back into the VDS

    We did all of this via the HTML5 portal of vcetner 6.5 (and of course ESX shell on the hosts).

    There wasn’t any reboots, no VMs were disconnected.

  3. it wasn’t hardware related after all, but I think the /tmp folder full didn’t help the hosts to complete the migration into new vcenter properly

Leave a Reply

Your email address will not be published. Required fields are marked *

Close