[ovirt-users] "wildcard resolve to multiple address" during hosted-engine --deploy

[Yedidyah Bar David]

(Adding the mailing list)

On Wed, Mar 16, 2016 at 4:14 AM, Zach <zfnoctis at gmail.com> wrote:
> Hi Didi,
>
> I chose to go with option #2, and I now am getting:
>
> [ ERROR ] Failed to execute stage 'Misc configuration': Failed to setup
> networks {'ovirtmgmt': {'nic': 'em1', 'bootproto': 'dhcp',
> 'blockingdhcp': True, 'defaultRoute': True}}. Error code: "29" message:
> "Determining IP information for ovirtmgmt... failed; no link present.
> Check cable?"

Please post setup logs.

>
> Which is *completely* nonsensical, as em1 is showing up as working both
> in CentOS, and iDrac - not to mention I am connected over ssh with that
> very same network interface.

Perhaps you do not have an IP address on that interface? You do need one.
The engine will communicate with the host using this address.

>  Just to be on the safe-side, I will
> mention that I disabled firewalld and Networkmanager as soon as I first
> logged into this machine. which also makes wonder why these need to be
> disabled, since they are included by default with CentOS/RHEL 7.
>
> This whole process has left me fairly frustrated and confused.

Sorry for that. Bugs/RFEs/patches are always welcome :-)

>  Why does
> a previous attempt somehow completely destroy any chance of installing
> the system without a complete wipe of the disks?

Because of the way the deploy script is currently implemented, it would be
hard to completely undo changes on failure (which we do try hard in
engine-setup). We have an open RFE for supplying a cleanup tool:

https://bugzilla.redhat.com/show_bug.cgi?id=1001181

See also:

http://www.ovirt.org/documentation/how-to/hosted-engine/#recoving-from-failed-install

The source package also has:
https://gerrit.ovirt.org/gitweb?p=ovirt-hosted-engine-setup.git;a=blob;f=cleanup.sh;h=89b315339022eff341042dd47d42bd979cf67176;hb=HEAD
It's currently down for some reason, not sure. A mirror is available at:
https://github.com/oVirt/ovirt-hosted-engine-setup/blob/master/cleanup.sh

All of these are far from perfect, and reinstalling the OS is way safer
when trying to debug such issues.

>  Why is the
> installation failing to even find a network interface? Is this par for
> the course to get oVirt installed?

Again, not sure what the exact problem was. Please post setup logs.

>
> Finally, are there any current guides on setting up oVirt 3.6 in a
> hosted-engine configuration with nfs/gluster all on a single machine?

Not that I am aware of.

loopback-mounting nfs is considered unsafe due to locking issues, see:
https://lwn.net/Articles/595652/

You can use iSCSI.

Most people install it on more than one machine. For a single machine,
usually virt-manager is enough.

Of course you can use VMs with nested-kvm to virtualize more than one
machine on a single one, if you mainly do this for testing/learning/etc.

> It looks like everything is old or incomplete.  But perhaps it is not
> intended to be used yet.

Proper support is currently partial. Search the net for e.g. 'ovirt
hyperconvergence' for the current efforts. The plan is to require at
least 3 physical machines, to have a replica-3 gluster storage and
quorum. It should be possible today to deploy on a single machine with
gluster, supplying three different paths for gluster all on the same
machine.

For best results, currently, you should have at least 3 machines:
One for storage (or a dedicated storage appliance), two for hosted-engine
hosts providing high-availability.

Best regards,

>
>
> On 03/14/2016 01:41 AM, Yedidyah Bar David wrote:
>> On Mon, Mar 14, 2016 at 3:19 AM, Zach <zfnoctis at gmail.com> wrote:
>>> I keep running into the following error when going through the
>>> "hosted-engine --deploy" configuration.
>>>
>>>     [ ERROR ] Failed to execute stage 'Setup validation': wildcard
>>> resolved to multiple address
>>>
>>> So far, I haven't been able to figure out what I'm doing wrong.  My best
>>> guess is that there is some issue with the FQDN that I used, but I'm not
>>> sure what I've done wrong.  I've included the following line in
>>> /etc/hosts for the FQDN.
>>>
>>>     192.168.0.50    ovirt.domain.local
>>>
>>> I've also attached the output log, as well as the full output of me
>>> answering all the prompts.
>>>
>>> Any help would be appreciated, let me know if I need to include more
>>> information.
>> It seems like there are some left-overs in vdsm from previous attempts.
>> If you want to further debug this, please:
>> 1. Try to give exact flow, including what you tried, changed etc at each point
>> 2. Attach all setup logs and all vdsm logs
>>
>> If you just want it all to work, please reinstall the machine from scratch
>> and try again, and if you get into problems on first attempt, check/post
>> setup and vdsm logs.
>>
>> Best,
>



-- 
Didi