From: "Daniel Helgenberger"
<daniel.helgenberger(a)m-box.de>
To: "Doron Fediuck" <dfediuck(a)redhat.com>
Cc: "Ryan Groten" <Ryan.Groten(a)stantec.com>, users(a)ovirt.org
Sent: Tuesday, September 23, 2014 2:06:39 PM
Subject: Re: [ovirt-users] How to disconnect hosted-engine NFS storage pool?
On 23.09.2014 11:51, Doron Fediuck wrote:
>
> ----- Original Message -----
>> From: "Daniel Helgenberger" <daniel.helgenberger(a)m-box.de>
>> To: "Doron Fediuck" <dfediuck(a)redhat.com>, "Ryan
Groten"
>> <Ryan.Groten(a)stantec.com>
>> Cc: users(a)ovirt.org
>> Sent: Tuesday, September 23, 2014 12:07:12 PM
>> Subject: Re: [ovirt-users] How to disconnect hosted-engine NFS storage
>> pool?
>>
>> Hello Ryan and Doron,
>>
>> Indeed a useful feature, I ran into this issue but forgot to mail. I
>> think in reality it is already best practice to put hosted engine on its
>> own NFS share (iSCSI LUN, Gluster-Vol in the future). Should we make
>> this an requirement?
>>
>>
>> On 23.09.2014 08:02, Doron Fediuck wrote:
>>> Ryan- perfect :)
>>> We'll take it from here.
>>>
>>> Thanks,
>>> Doron
>>>
>>> ----- Original Message -----
>>>> From: "Ryan Groten" <Ryan.Groten(a)stantec.com>
>>>> To: "Doron Fediuck" <dfediuck(a)redhat.com>
>>>> Cc: users(a)ovirt.org
>>>> Sent: Monday, September 22, 2014 8:03:30 PM
>>>> Subject: RE: [ovirt-users] How to disconnect hosted-engine NFS storage
>>>> pool?
>>>>
>>>> Thanks, RFE created (I hope I did it right)
>>>>
https://bugzilla.redhat.com/show_bug.cgi?id=1145259
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: Doron Fediuck [mailto:dfediuck@redhat.com]
>>>> Sent: September-21-14 6:57 AM
>>>> To: Groten, Ryan
>>>> Cc: users(a)ovirt.org
>>>> Subject: Re: [ovirt-users] How to disconnect hosted-engine NFS storage
>>>> pool?
>>>>
>>>>
>>>>
>>>> ----- Original Message -----
>>>>> From: "Ryan Groten" <Ryan.Groten(a)stantec.com>
>>>>> To: users(a)ovirt.org
>>>>> Sent: Friday, September 19, 2014 1:51:13 AM
>>>>> Subject: [ovirt-users] How to disconnect hosted-engine NFS storage
>>>>> pool?
>>>>>
>>>>>
>>>>>
>>>>> I want to unmounted the hosted-engine NFS share without affecting
all
>>>>> the other running VMs on the host. When I shutdown the
hosted-engine
>>>>> and enable global maintenance, the storage pool is still mounted and
I
>>>>> can’t unmount it because the “sanlock” process is using it.
>>>>>
>>>>>
>>>>>
>>>>> Is there any way to disconnect the storage pool? There is a
>>>>> hosted-engine --connect-storage option but I see nothing to
disconnect
>>>>> it.
>>>>>
>>>>>
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Ryan
>>>>>
>>>> Hi Ryan,
>>>> Hosted engine does not unmount the share since there may be other VMs
>>>> using
>>>> it (as a general rule).
>>>> However this may deserve some additional thoughts. Do you mind opening
>>>> an
>>>> RFE
>>>> for it?
>>>>
>>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users(a)ovirt.org
>>>
http://lists.ovirt.org/mailman/listinfo/users
>>>
>> --
>> Daniel Helgenberger
>> m box bewegtbild GmbH
>>
>> P: +49/30/2408781-22
>> F: +49/30/2408781-10
>>
>> ACKERSTR. 19
>> D-10115 BERLIN
>>
>>
>>
www.m-box.de www.monkeymen.tv
>>
>> Geschäftsführer: Martin Retschitzegger / Michaela Göllner
>> Handeslregister: Amtsgericht Charlottenburg / HRB 112767
>>
>>
> Hi Daniel,
> we're trying to reduce the special features hosted engine has.
> The ideal is to have it as every other VM, so I'd rather not
> add requirements.
Hello Doron,
makes absolutely sense (at least for developers who need to test things
a lot ;)
From a practical point of view though it will always a bit different as
long as you do not plan major design changes with hosted-engine because
of the potential lockout problem (engine-ha and shutdown / reboot, IIRC
right now no management other then migration from within ovirt). I know
you will have engine more 'manageable' form oVirt itself in the future,
but for the following reasons this is not very important for me:
Since in a worst case scenario engine is crustal getting tings working
again, 'users' (at least me) would tend to use a much simpler/robust
setup for engine storage (for instance, plain nfs, gluster 'local'
cluster - one replicated brick per host, no dm-multipath ect.). As for
the storage of 'production' VMs other points tend to be more crucial
(performance, multiple pathes, easy provision and administration ...).
My point being, common sense suggests you setup engine storage different
from 'production' storage.
Daniel
>
>
> Doron
>
--
Daniel Helgenberger
m box bewegtbild GmbH
P: +49/30/2408781-22
F: +49/30/2408781-10
ACKERSTR. 19
D-10115 BERLIN
www.m-box.de www.monkeymen.tv
Geschäftsführer: Martin Retschitzegger / Michaela Göllner
Handeslregister: Amtsgericht Charlottenburg / HRB 112767
Daniel,
keeping the engine VM as close to the others is not only done
from the developer perspective, but rather from an admin perspective
as well. ie- try to keep things simple as much as possible.
Every admin may have his own way of managing the system, and
this is perfectly fine. So taking precaution by having a special
NFS share makes sense, and would be a good best practice.
Thanks for the feedback!
Doron