[ovirt-users] VM Migration Fails (Network thoughts)

[Christopher Young]

Annnddd.... here I am again being 'that guy'.  I apologize for all the
traffic, but I wanted to note that migrations appears to be working
even over the migration IP network once short name resolution was
working fine.  The first attempt did fail after changes, but since
then, they are working fine, it seems.

So, I'm down to only a simple question now:

If I edit the resolv.conf on a RHEV-H/ovirt node (7.2) host, does it
REALLY get replaced on reboot?  And if so, how do I work around this
since I need those name resolutions to work?  Would /etc/hosts file
entries be sufficient for just the primary hostnames and IPs (or would
they also get over-written on reboot)?

Again, sorry for unnecessary posts, but I'm just trying to understand
the pieces as much as possible since I'm going to be supporting this
for a very long time.

On Fri, Mar 4, 2016 at 3:42 PM, Christopher Young <mexigabacho at gmail.com> wrote:
> So, I'm self-replying, but I was able to confirm that changing
> /etc/resolv.conf so that the short hostnames work allow migration, but
> ONLY when the migration network is set to the main management
> interface (and thus the primary hostname of the system).  I'm
> wondering how this could work with separate IP networks for VM
> Migration when hostname resolutions would generally always return the
> primary IP address of a system.
>
> I obviously still have reading to do, but any help and feedback would
> be greatly appreciated.
>
>
>
> On Fri, Mar 4, 2016 at 3:31 PM, Christopher Young <mexigabacho at gmail.com> wrote:
>> So,  I'm attempting to understand something fully.  My setup might not
>> be ideal, so please provide me whatever education I may need.
>>
>> RE: https://access.redhat.com/solutions/202823
>>
>> Quick background:
>>
>> I have (3) RHEV-H (ovirt node) Hypervisors with a hosted-engine.  VM
>> Migrations currently are not working and I have the following errors
>> in the vdsm.log:
>>
>> ---
>> "[RHEV]Failed to migrate VM between hypervisor with error "gaierror:
>> [Errno -2] Name or service not known" in vdsm.log"
>> ---
>>
>> Reading the RedHat article leads me to believe that this is a name
>> resolution issue (for the FQDNs), but that leads to another question
>> which I'll get to.  First, a little more about the setup:
>>
>> I have separate IP networks for Management (ovirtmgmt), Storage, and
>> VM Migration.  These are all on unique IP ranges.  I'll make up some
>> for my purposes here:
>>
>> MGMT_VLAN: 10.25.250.x/24
>> STORAGE_VLAN: 10.26.3.x/24
>> MIGRATE_VLAN: 10.26.5.x/24
>>
>> In anticipation for setting all of this up, I assigned hostnames/IPs
>> for each role on each node/hypervisor, like so:
>>
>> vnode01
>> vnode01-sto (storage)
>> vnode01-vmm
>>
>> vnode02
>> vnode02-sto
>> vnode02-vmm
>>
>> etc., etc.
>>
>> So, I'm trying to understand what my best procedure for setting up a
>> separate migration network and what IP/hostname settings are necessary
>> to ensure that this works properly and wouldn't affect other things.
>>
>> I'm going to do some more reading right now, but at least I feel like
>> I'm on the right path to getting migrations working.
>>
>> Please help lol
>>
>> Thanks as always,
>>
>> -- Chris