11:44 a.m.
Hi Fran,
thanks for helping me.
I have defined a separate fqdn and ip address for the engine - sorry for
not mentioning that. It is also in /etc/hosts.
The problem seems to be the following:
The error message is thrown by ansible if
he_host_ip not in target_address_v4.stdout_lines and
he_host_ip not in target_address_v6.stdout_lines
and grepping he_host_ip from
ovirt-hosted-engine-setup-ansible-initial_clean-XYZ.log turns out to be
the ip address of the physical interface but target_address_v4 is
explicitly set to "ip addr show" of the vlan interface.
Since I defined the vlan interface as bridge interface in the VM
settings for the engine this seems strange to me.
What is more strange: If I put in the physical host interface as bridge
interface the error does not occur, but then the Engine-VM is bound to
the default libvirt bridge (which I did not want it to) and the setup
process fails with "There was a failure deploying the engine on the
local engine VM." which I now have to analyze.
Is my idea of binding the bridge to the vlan interface to have a
management network there completely wrong?
I did not manage to find any docs on what the cockpit module expects
there, and the ovirt setup docs are also very sparse - can you point me
to some in depth examples of the requirements the self hosted engine
setup expects?
Cheers,
Dirk
Am 21.03.23 um 15:30 schrieb Fran Garcia:
The hosted engine is a regular Virtual machine.
You need to have a separate FQDN and IP, different from those assgined
to the hypervisors.
In your example, you need to assign a fqdn/IP from the vlan 4000 (or
wherever the mgmt vlan is), and use it whenever queried about the
Hosted Engine VM details.
HTH
Fran
On Tue, 21 Mar 2023 at 13:07, <dirk.schulz(a)geniodata.de> wrote:
> Hi all,
> I have set up 3 servers in 3 data centers, each having one physical interface and a
vlan interface parented by it.
> The connection between the 3 servers over the vlan interfaces (using private ip
addresses) works (using icmp ping as the test).
>
> Now I want to turn them into an ovirt cluster creating the self hosted engine on the
first server. I have
> - made sure the engine fqdn is in dns forward and reverse and in /etc/hosts
> - made sure that both interfaces have unique dns entries which can be resolved
forward and reverse
> - made sure that both interfaces' fqdns are in /etc/hosts
> - made sure only the primary hostname (not fqdn) is in /etc/hostname,
> - made sure ipv6 is available on the physical interface,
> - made sure ipv6 method is "disabled" on the vlan interface,
> - set
/usr/share/ansible/collections/ansible_collections/ovirt/ovirt/roles/hosted_engine_setup/defaults/main.yml:he_force_ip4:
true to make sure no ipv6 attempts to interfere.
>
> Now when I use cockpit's hosted engine wizard (not hyperconverged), i run into 2
opposing problems.
> If I set the FQDN in the "Advanced" sub pane to the FQDN of the vlan
interface, the wizards gets stuck at "preparing VM" with "The resolved
address doesn't resolve on the selected interface\n".
> If I set the FQDN in the "Advanced" sub pane to the FQDN of the physical
interface, I get the same result.
>
> If i add the physical interfaces FQDN to the vlan ip address in /etc/hosts, i get
"hostname 'x.y.z' doesn't uniquely match the interface
'enp5s0.4000' selected for the management bridge; it matches also interface with
IP ['physical']. Please make sure that the hostname got from the interface for the
management network resolves only there." So clearly separating the two interfaces
namewise is mandatory.
>
> I tried to follow the ansible workflow step by step to see what it does. I seems the
validate hostname is triggered twice, second time on filling in FQDN in
"Advanced" sub pane - it succeeds with both hostnames (physiscal interface and
vlan ip), but that does not prevent the "prepare VM" workflow in doing the same
verification and failing, as far as I can see. This is where it happens:
> 2023-03-20 14:31:48,354+0100 DEBUG ansible on_any args TASK:
ovirt.ovirt.hosted_engine_setup : Check the resolved address resolves on the selected
interface kwargs is_conditional:False
> 2023-03-20 14:31:48,355+0100 DEBUG ansible on_any args localhost TASK:
ovirt.ovirt.hosted_engine_setup : Check the resolved address resolves on the selected
interface kwargs
> 2023-03-20 14:31:48,481+0100 DEBUG var changed: host "localhost" var
"ansible_play_hosts" type "<class 'list'>" value:
"[]"
> 2023-03-20 14:31:48,481+0100 DEBUG var changed: host "localhost" var
"ansible_play_batch" type "<class 'list'>" value:
"[]"
> 2023-03-20 14:31:48,481+0100 DEBUG var changed: host "localhost" var
"play_hosts" type "<class 'list'>" value:
"[]"
> 2023-03-20 14:31:48,481+0100 ERROR ansible failed {
> "ansible_host": "localhost",
> "ansible_playbook":
"/usr/share/ovirt-hosted-engine-setup/ansible/trigger_role.yml",
> "ansible_result": {
> "_ansible_no_log": false,
> "changed": false,
> "msg": "The resolved address doesn't resolve on the
selected interface\n"
> },
> "ansible_task": "Check the resolved address resolves on the
selected interface",
> "ansible_type": "task",
> "status": "FAILED",
> "task_duration": 0
> }
>
>
> So I am really stuck there. I do not have any idea how and where to go on. I can try
changing bits in the playbooks and parameters (like using "hostname -A" instead
of "hostname -f" for the failing test), but that can't really be the idea -
I am to new to this to run into a bug or similar, I will suspect I do overlook something.
>
> Any hint or help is appreciated.
>
> Cheers,
>
> Dirk
> _______________________________________________
> Users mailing list -- users(a)ovirt.org
> To unsubscribe send an email to users-leave(a)ovirt.org
> Privacy Statement: https://www.ovirt.org/privacy-policy.html
> oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/
> List Archives:
https://lists.ovirt.org/archives/list/users@ovirt.org/message/X2DVC2Q22SI...