On Fri, 2013-11-08 at 08:49 +0100, Nicolas Ecarnot wrote:
Le 08/11/2013 08:23, Eli Mesika a écrit :
>
>
> ----- Original Message -----
>> From: "Mike Burns" <mburns(a)redhat.com>
>> To: "Nicolas Ecarnot" <nicolas(a)ecarnot.net>, users(a)ovirt.org
>> Sent: Thursday, November 7, 2013 4:10:36 PM
>> Subject: Re: [Users] Fencing virtual machines
>>
>> On 11/07/2013 09:03 AM, Nicolas Ecarnot wrote:
>>> Hi,
>>>
>>> In the case I want to setup a virtual cluster amongst many virtual
>>> machines, and this virtual cluster needs to fence some VM, what is the
>>> best way to achieve this?
>>>
>>> I guess this has to use something like vdsClient to "unplug" the
VM, but
>>> I found it hard to find any more information about this.
>>>
>>
>> I haven't tried but the fence-agents package (fence-agents-rhevm on
>> Fedora) has a fence_rhevm tool that will probably work for you.
>
> Please let us know if that worked for you.
Thank you all for your answer. I installed it and began to try to play
with it, and it seems to be EXACTLY what I was looking for.
It's still not clear for me to see what I have to install on the
"stonith" targets - or who is eventually in charge of applying the kill
action (the manager, one node, any node ?) but I promise I'm still
reading and testing on my own.
Your vms are killed/restarted by ovirt-engine using the REST-API (on
behalf of one of the cluster members of course). So you don't have to
install anything except the fence-agents on all your cluster members
(vms) and make sure they can reach port 443 (or 8443 depending on your
setup) of ovirt-engine...
Regards,
René
Thank you, have a nice day.