On 29 Mar 2019, at 14:42, Callum Smith <callum@well.ox.ac.uk> wrote:

Ok so we're getting very close now, weird OVF error:

Full ansible log attached

Only error in the engine.log looks normal/expected to me:

2019-03-29 14:32:44,370Z ERROR [org.ovirt.engine.core.bll.pm.FenceProxyLocator] (EE-ManagedThreadFactory-engineScheduled-Thread-71) [4405d6db] Can not run fence action on host 'vir

thyp04.virt.in.bmrc.ox.ac.uk', no suitable proxy host was found.

<ovirt.20190329133000.ansible.log>

Feeling damn close to success here, but have managed to replicate this issue twice re-running the installer.

Regards,

Callum

--

Callum Smith

Research Computing Core

Wellcome Trust Centre for Human Genetics
University of Oxford
e. callum@well.ox.ac.uk

On 29 Mar 2019, at 11:50, Simone Tiraboschi <stirabos@redhat.com> wrote:

On Fri, Mar 29, 2019 at 12:36 PM Callum Smith <callum@well.ox.ac.uk> wrote:

ip link del ovirtmgmt has done the job....

Another issue, but this is likely due to randomised MAC addresses:

fatal: [virthyp04.virt.in.bmrc.ox.ac.uk]: FAILED! => {"changed": true, "cmd": ["virt-install", "-n", "HostedEngineLocal", "--os-variant", "rhel7", "--virt-type", "kvm", "--memory", "4096", "--vcpus", "64", "--network", "network=default,mac=fe:58:6c:da:1e:cc,model=virtio", "--disk", "/var/tmp/localvmOCYiyF/images/6f76686b-199c-4cb3-bbbe-86fc34365745/72bc3948-5d8d-4877-bac8-7db4995045b5", "--import", "--disk", "path=/var/tmp/localvmOCYiyF/seed.iso,device=cdrom", "--noautoconsole", "--rng", "/dev/random", "--graphics", "vnc", "--video", "vga", "--sound", "none", "--controller", "usb,model=none", "--memballoon", "none", "--boot", "hd,menu=off", "--clock", "kvmclock_present=yes"], "delta": "0:00:01.355834", "end": "2019-03-29 11:31:02.100143", "msg": "non-zero return code", "rc": 1, "start": "2019-03-29 11:31:00.744309", "stderr": "ERROR    unsupported configuration: Unable to use MAC address starting with reserved value 0xFE - 'fe:58:6c:da:1e:cc' - \nDomain installation does not appear to have been successful.\nIf it was, you can restart your domain by running:\n  virsh --connect qemu:///system start HostedEngineLocal\notherwise, please restart your installation.", "stderr_lines": ["ERROR    unsupported configuration: Unable to use MAC address starting with reserved value 0xFE - 'fe:58:6c:da:1e:cc' - ", "Domain installation does not appear to have been successful.", "If it was, you can restart your domain by running:", "  virsh --connect qemu:///system start HostedEngineLocal", "otherwise, please restart your installation."], "stdout": "\nStarting install...", "stdout_lines": ["", "Starting install..."]}

Seems it doesn't take into account reserved values when generating.

If not specified by the user,

a random unicast MAC address is randomly generated here:

https://github.com/oVirt/ovirt-ansible-hosted-engine-setup/blob/master/tasks/pre_checks/validate_mac_address.yml#L4

for sure it's unicast but maybe we should make it more robust on reserved values.

Simply try again for now; I'll open an issue to track it, thanks!

 

I hope this feedback is valuable - I have a good feeling about the current deploy otherwise though.

Regards,

Callum

--

Callum Smith

Research Computing Core

Wellcome Trust Centre for Human Genetics
University of Oxford
e. callum@well.ox.ac.uk

On 29 Mar 2019, at 11:01, Simone Tiraboschi <stirabos@redhat.com> wrote:

On Fri, Mar 29, 2019 at 11:56 AM Callum Smith <callum@well.ox.ac.uk> wrote:

Dear Simone,

It doesn't seem to want to work:

# Settings

he_fqdn: "he.virt.in.bmrc.ox.ac.uk"

he_ansible_host_name: "virthyp04.virt.in.bmrc.ox.ac.uk"

he_admin_password:  <snip>

he_appliance_password:  <snip>

# Resources

he_mem_size_MB: "4096"

# Storage

he_domain_type: "nfs"

he_storage_domain_addr: <snip>

he_storage_domain_path: <snip>

# Network

he_vm_ip_addr: "10.141.31.240"

he_vm_ip_prefix: "20"

he_dns_addr: ["10.141.31.251","10.141.31.252","10.141.31.253"]

he_default_gateway_4: "10.141.31.254"

he_gateway: he_default_gateway_4

he_force_ip4: true

he_bridge_if: bond0.910

#he_just_collect_network_interfaces: true

# Email

he_smtp_port: 25

he_smtp_server: smtp.ox.ac.uk

he_dest_email: rescomp-ops@well.ox.ac.uk

he_source_email: ovirt@bmrc.ox.ac.uk

# Ansible Stuff

ansible_ssh_user: root

ansible_become: false

host_key_checking: false

I've attached the output of the ansible command as a log file, this is what happens when the IF bond0.910 is assigned the IP and `ovirtmgmt` is not defined on the host.

TASK [ovirt.hosted_engine_setup : debug] *******************************************************************************************************************************************

ok: [virthyp04.virt.in.bmrc.ox.ac.uk] => {

    "target_address_v4": {

        "changed": true, 

        "cmd": "ip addr show ovirtmgmt | grep 'inet ' | cut -d' ' -f6 | cut -d'/' -f1", 

        "delta": "0:00:00.008744", 

        "end": "2019-03-29 10:26:07.510481", 

        "failed": false, 

        "rc": 0, 

        "start": "2019-03-29 10:26:07.501737", 

        "stderr": "", 

        "stderr_lines": [], 

        "stdout": "", 

        "stdout_lines": []

    }

}

according to the logs ovirtmgmt is still there.

can you please share the output of 'ip a'?

 

Regards,

Callum

--

Callum Smith

Research Computing Core

Wellcome Trust Centre for Human Genetics
University of Oxford
e. callum@well.ox.ac.uk

On 28 Mar 2019, at 16:23, Simone Tiraboschi <stirabos@redhat.com> wrote:

On Thu, Mar 28, 2019 at 1:44 PM Callum Smith <callum@well.ox.ac.uk> wrote:

Dear Simone,

This is my experience too, but I'm now hitting this error on the hosted-engine install at the part where it registers the hypervisor as the first host in the engine:

2019-03-28 12:40:50,025Z INFO  [org.ovirt.engine.core.bll.host.HostConnectivityChecker] (EE-ManagedThreadFactory-engine-Thread-1) [49f371c1] Engine managed to communicate with VDSM

 agent on host 'virthyp04.virt.in.bmrc.ox.ac.uk' with address 'virthyp04.virt.in.bmrc.ox.ac.uk' ('db571f8a-fc85-40d3-b86f-c0038e3cd7e7')

2019-03-28 12:40:53,111Z WARN  [org.ovirt.engine.core.bll.network.NetworkConfigurator] (EE-ManagedThreadFactory-engine-Thread-1) [49f371c1] Failed to find a valid interface for the

 management network of host virthyp04.virt.in.bmrc.ox.ac.uk. If the interface ovirtmgmt is a bridge, it should be torn-down manually.

2019-03-28 12:40:53,111Z ERROR [org.ovirt.engine.core.bll.hostdeploy.InstallVdsInternalCommand] (EE-ManagedThreadFactory-engine-Thread-1) [49f371c1] Exception: org.ovirt.engine.cor

e.bll.network.NetworkConfigurator$NetworkConfiguratorException: Interface ovirtmgmt is invalid for management network

The host's ovirtmgmt network connection is a statically assigned IP on a VLAN on a bond, how should I be configuring this if not manually?

If you need to deploy over vlan 123 over bond0 simply configure a device exactly called bond0.123 and statically configure your IP address there.

Choose it for hosted-engine deployment, nothing more: ovirtmgmt will be automatically created over that device and the vlan ID will be set at engine level for the whole management network.

 

Regards,

Callum

--

Callum Smith

Research Computing Core

Wellcome Trust Centre for Human Genetics
University of Oxford
e. callum@well.ox.ac.uk

On 27 Mar 2019, at 17:09, Simone Tiraboschi <stirabos@redhat.com> wrote:

On Wed, Mar 27, 2019 at 4:27 PM Callum Smith <callum@well.ox.ac.uk> wrote:

It's ok, migrating to 4.3.2 on the oVirt node (from 4.3.0) did the job of fixing it.

It is a bug if you intend on using the ovirtmgmt network to deploy your ansible from

This is a bit tricky: when the engine brings up the host it also creates the management bridge and this could lead to a temporary network down on the selected interface for the bridge creation time (a couple of seconds?)

I tried it on a LAN and ansible ssh connection always survived but I 'm not sure it's always true.

 

, and you need it to have an IP address already on that range! But - it works as expected with the ovirtmgmt bridge setup so nothing to worry about.

Regards,

Callum

--

Callum Smith

Research Computing Core

Wellcome Trust Centre for Human Genetics
University of Oxford
e. callum@well.ox.ac.uk

On 27 Mar 2019, at 14:57, Simone Tiraboschi <stirabos@redhat.com> wrote:

On Wed, Mar 27, 2019 at 3:24 PM Callum Smith <callum@well.ox.ac.uk> wrote:

Dear All,

We're trying to deploy our hosted engine remotely using the ansible hosted engine playbook, which has been a rocky road but we're now at the point where it's installing, and failing. We've got a pre-defined bond/VLAN setup for our interface which has the correct bond0 bond0.123 and ovirtmgmt bridge on top but we're hitting the classic error:

Failed to find a valid in

terface for the management network of host virthyp04.virt.in.bmrc.ox.ac.uk. If the interface ovirtmgmt is a bridge, it should be torn-down manually.

Does this bug still exist in the latest (4.3) version, and is installing using ansible with this network configuration impossible?

I don't think it's a bug; please avoid manually creating ovirtmgmt and simply set:
he_bridge_if: "bond0.123", in ansible variable file and the management bridge will be created for you at host-deploy time.

 

Regards,

Callum

--

Callum Smith

Research Computing Core

Wellcome Trust Centre for Human Genetics
University of Oxford
e. callum@well.ox.ac.uk

_______________________________________________
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-leave@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/
List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/SBOZ6FRBRQKE7FMTEHT6FBKHOW3PIPHK/