Re: [ovirt-users] Hi everyone, new guy in the mail list
On Thu, May 25, 2017 at 11:01 AM Yaniv Kaul <ykaul(a)redhat.com> wrote:
On Wed, May 24, 2017 at 9:28 PM, Odilon Junior
<odilon.junior93(a)gmail.com>
wrote:
> Hi everyone, i've just subscribed to the maling list, I was going to ask
> one thing about ovirt in the IRC channel but no one was there.
>
> I've been using ovirt since 3.1, today we have one enviroment with 1
> Ovirt Engine and 3 baremetal that we use for virtualization running the
> Ovirt 3.6.3. We have 3 NFS share from our colocation provider that are used
> as Storage.
>
> Our collocation will be rolling one emergency maintenance to upgrade the
> firmware on the routers that provides our Private Network, the NFS service
> run on this Private Network,the downtime should be no more than 3 minutes.
>
> I'm concerned, what is the best way to approach this, should we shutdown
> all the VMs and the hypervisor? Or we can just let the Ovirt handle this?
> The communication between the Ovirts and the Engine are from another
> network that will not be affected.
>
You should put everything into maintenance.
Unlike other storages, NFS will hang and you'll have stale mounts
eventually, which cannot be recovered from.
We have code detecting and recovering from stale mounts, but we also
have a bug reporting stale mounts that did not recover.
If the vms using this storage are very important, and you cannot have
downtime, best move the disk to another storage before the upgrade,
and move back the disk to the original storage after the upgrade.
If the vms are not very important and you are ok with several minutes
of downtime, stopping the vms and putting the domain to maintenance
is the safe way.
Nir
Attachments:
- attachment.html (text/html — 2.5 KB)