[ovirt-users] Hosted engine deploy: problem with the rDNS of another iface

[Simone Tiraboschi]

On Wed, Sep 14, 2016 at 2:53 PM, Davide Ferrari <davide at billymob.com> wrote:

> And indeed it was the problem. So, you must have working reverse dns
> lookup to complete the setup succesfully.
>

Yes, this is a requirement to avoid other issues.


>
> Thanks for the help!
>
> Ciao!
>
> 2016-09-14 14:48 GMT+02:00 Davide Ferrari <davide at billymob.com>:
>
>> Well but the fact is that it DOESN'T resolve on bond1, there's no DNS
>> entry anywhere resolving to that IP for vm01.mydomain.tld
>> That's the strange thing, I don't really get from where python's
>> socket.gethostbyaddr() gets those 2 IP addresses :/
>>
>> The only thing that it comes to my mind is that the domain I'm using
>> (mydomain.tld) doesn't have PTRs properly set (and I cannot fix it without
>> migrating the domain...yeah it's ugly as hell what I have here, I know :(
>> ), maybe it's a corner case that should be handled more specifically?
>>
>> I'll try to add entries for vm01.mydomain.tld in /etc/hosts and see if it
>> fixes the problem
>>
>> 2016-09-14 14:32 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>
>>>
>>>
>>> On Wed, Sep 14, 2016 at 2:07 PM, Davide Ferrari <davide at billymob.com>
>>> wrote:
>>>
>>>> But bond0/192.168.10.225 is the only interface I have (right now) to
>>>> connect to the machine. So, I should create another interface, dedicated to
>>>> management? Right now I'm using bond0 as the source iface for the mgm
>>>> bridge.
>>>>
>>>
>>> No this is absolutely fine; the problem is just that 'vm01.mydomain.tld'
>>> should uniquely resolve there and it should not resolve also on bond1
>>> otherwise you can hit the issue I mentioned before.
>>>
>>>
>>>>
>>>> 2016-09-14 14:02 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>
>>>>>
>>>>>
>>>>> On Wed, Sep 14, 2016 at 1:53 PM, Davide Ferrari <davide at billymob.com>
>>>>> wrote:
>>>>>
>>>>>> Ciao Simone!
>>>>>>
>>>>>> Absolutely, here it is (please not that I've obfuscated domain names
>>>>>> but they are consistent with what I wrote before, the hostnames are exactly
>>>>>> the same I have in my env). If you need the original log, tell me but I
>>>>>> will mail it to you privately.
>>>>>>
>>>>>
>>>>> The issue is here:
>>>>>
>>>>> 2016-09-14 13:47:51 DEBUG otopi.plugins.gr_he_common.network.bridge
>>>>> bridge._get_hostname_from_bridge_if:325 hostname:
>>>>> 'vm01.mydomain.tld', aliaslist: '[]', ipaddrlist: '['192.168.10.225',
>>>>> '10.30.0.1']'
>>>>>
>>>>> You can reproduce it with:
>>>>>     python -c 'import socket; print(socket.gethostbyaddr("19
>>>>> 2.168.10.225"));'
>>>>>
>>>>> 'vm01.mydomain.tld' should resolve just on the interface you selected
>>>>> to create the management bridge on.
>>>>> This is because hosted-engine-setup is creating the management bridge
>>>>> on that interface before starting the engine VM and when when it will call
>>>>> hosts.add on the REST API to add your first host to the engine it can just
>>>>> specify an host address by not an interface name and so host-deploy,
>>>>> triggered by the engine, will recreate the management bridge on the
>>>>> interface where the address we passed resolves on.
>>>>> So we have to avoid any ambiguity here.
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>> Thanks!
>>>>>>
>>>>>> 2016-09-14 13:37 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>>
>>>>>>>
>>>>>>> On Wed, Sep 14, 2016 at 1:25 PM, Davide Ferrari <davide at billymob.com
>>>>>>> > wrote:
>>>>>>>
>>>>>>>> Hello
>>>>>>>>
>>>>>>>> I've got the same exact problem as in this thread:
>>>>>>>>
>>>>>>>> http://lists.ovirt.org/pipermail/users/2016-July/041111.html
>>>>>>>>
>>>>>>>> (sorry for not answering there but I've just subscribed). The error
>>>>>>>> (for the record) is:
>>>>>>>>
>>>>>>>> Failed to execute stage 'Setup validation': hostname
>>>>>>>> 'vm01.mydomain.tld' doesn't uniquely match the interface 'bond0' selected
>>>>>>>> for the management bridge; it matches also interface with IP
>>>>>>>> set(['10.30.0.1']). Please make sure that the hostname got from the
>>>>>>>> interface for the management network resolves only there.
>>>>>>>>
>>>>>>>> 10.30.0.1 is set on bond1 and is used for the glusterfs network.
>>>>>>>> The python output is:
>>>>>>>>
>>>>>>>> python -c 'import socket; print(socket.gethostbyaddr("10
>>>>>>>> .30.0.1"));'
>>>>>>>> ('glu-ovirt01.storage', [], ['10.30.0.1'])
>>>>>>>>
>>>>>>>> which obviously is completely different from vm01.mydomain.tld.
>>>>>>>> So, I don't really get this error. I've used
>>>>>>>> glu-ovirt01.storage:/engine as the glusterfs entry point for storing the
>>>>>>>> engine image, that's all.
>>>>>>>>
>>>>>>>> Any idea on how to fix the error?
>>>>>>>>
>>>>>>>
>>>>>>> Ciao Davide,
>>>>>>> can you please attach the whole hosted-engine-setup log file?
>>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>> TIA
>>>>>>>>
>>>>>>>> --
>>>>>>>> Davide Ferrari
>>>>>>>> Senior Systems Engineer
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Users mailing list
>>>>>>>> Users at ovirt.org
>>>>>>>> http://lists.ovirt.org/mailman/listinfo/users
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Davide Ferrari
>>>>>> Senior Systems Engineer
>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Davide Ferrari
>>>> Senior Systems Engineer
>>>>
>>>
>>>
>>
>>
>> --
>> Davide Ferrari
>> Senior Systems Engineer
>>
>
>
>
> --
> Davide Ferrari
> Senior Systems Engineer
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20160914/a682517e/attachment-0001.html>