Re: [ovirt-users] VMs HA with cinder volumes

[Matteo Dacrema]

Ok, it works.

Thank you very much Nir.

Matteo



> Il giorno 21 feb 2017, alle ore 18:26, Nir Soffer  ha 
> scritto:
> 
> On Tue, Feb 21, 2017 at 3:20 PM, Matteo Dacrema  wrote:
>> Hi Nir,
>> 
>> thank you.
>> So, can I use a Ceph backed VM with a lease on gluster?
> 
> Should work.
> 
>> Also if I detach network from the node hosting the VMs it will not migrated
>> to another host.
> 
> If you have a lease, detaching the network used to access the leases will
> cause sanlock to terminate the vm and expire the lease. In this case the
> system should start the vm on another host.
> 
> Note that if you have multiple networks, and the network used by the vm
> disks is different from the network used to access the leases, all this may
> not work in a useful way.
> 
> Either the leases will not be accessible while the vm disks are accessible,
> leading to killing a good working vm, or the vm disk are not accessible,
> pausing the vm, but since the leases are still accessible, the vm will not
> release the lease and the system will not be able to start the vm elsewhere.
> 
> Maybe you can expose some of the ceph storage using cephfs for the leases?
> This will make it more likely that the leases storage domain will fail
> when there
> is some issue with your ceph cluster.
> 
> Another option is to expose ceph block device using iscsi, and use this iscsi
> storage domain for the leases. But note that this makes your iscsi gateway
> host a single point on failure.
> 
> Nir
> ___
> Users mailing list
> Users@ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
> 
> --
> Questo messaggio e' stato analizzato con Libra ESVA ed e' risultato non 
> infetto.
> Seguire il link qui sotto per segnalarlo come spam: 
> http://mx01.enter.it/cgi-bin/learn-msg.cgi?id=D2CE14011D.A4808
> 
> 

___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

Re: [ovirt-users] VMs HA with cinder volumes

[Nir Soffer]

On Tue, Feb 21, 2017 at 3:20 PM, Matteo Dacrema  wrote:
> Hi Nir,
>
> thank you.
> So, can I use a Ceph backed VM with a lease on gluster?

Should work.

> Also if I detach network from the node hosting the VMs it will not migrated
> to another host.

If you have a lease, detaching the network used to access the leases will
cause sanlock to terminate the vm and expire the lease. In this case the
system should start the vm on another host.

Note that if you have multiple networks, and the network used by the vm
disks is different from the network used to access the leases, all this may
not work in a useful way.

Either the leases will not be accessible while the vm disks are accessible,
leading to killing a good working vm, or the vm disk are not accessible,
pausing the vm, but since the leases are still accessible, the vm will not
release the lease and the system will not be able to start the vm elsewhere.

Maybe you can expose some of the ceph storage using cephfs for the leases?
This will make it more likely that the leases storage domain will fail
when there
is some issue with your ceph cluster.

Another option is to expose ceph block device using iscsi, and use this iscsi
storage domain for the leases. But note that this makes your iscsi gateway
host a single point on failure.

Nir
___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

Re: [ovirt-users] VMs HA with cinder volumes

[Matteo Dacrema]

Hi Nir,

thank you.
So, can I use a Ceph backed VM with a lease on gluster?

Also if I detach network from the node hosting the VMs it will not migrated to 
another host.


Regards
Matteo



> Il giorno 20 feb 2017, alle ore 21:58, Nir Soffer  ha 
> scritto:
> 
> On Mon, Feb 20, 2017 at 12:54 PM, Matteo Dacrema  > wrote:
> Hi all,
> 
> I’ve a setup that uses Ceph through ovirt for VM storage.
> I tested out VM HA and all works well when IPMI is reachable.
> When it’s not reachable because of network or power failure VM states switch 
> to unknown and it’s not restarted to other nodes.
> 
> Is it a specific setup that I’ve missed?
> 
> Hi Matteo,
> 
> When power management is not available, we cannot do failover to other host.
> 
> In 4.1 you can use a VM lease for your HA VM, enabling failover without power
> management.
> 
> Note that the lease cannot be on Ceph storage domain, you will have to add
> other type of storage for the leases, since we don't support leases on ceph 
> yet.
> 
> Please check the vm-leases feature page for more info:
> https://www.ovirt.org/develop/release-management/features/storage/vm-leases/ 
> 
> 
> Nir
>  
> 
> Thank you
> Regards
> Matteo
> 
> ___
> Users mailing list
> Users@ovirt.org 
> http://lists.ovirt.org/mailman/listinfo/users 
> 
> 
> 
> -- 
> Questo messaggio e' stato analizzato con Libra ESVA ed e' risultato non 
> infetto. 
> Clicca qui per segnalarlo come spam. 
>  
> Clicca qui per metterlo in blacklist 
> 
___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

Re: [ovirt-users] VMs HA with cinder volumes

[Nir Soffer]

On Mon, Feb 20, 2017 at 12:54 PM, Matteo Dacrema  wrote:

> Hi all,
>
> I’ve a setup that uses Ceph through ovirt for VM storage.
> I tested out VM HA and all works well when IPMI is reachable.
> When it’s not reachable because of network or power failure VM states
> switch to unknown and it’s not restarted to other nodes.
>
> Is it a specific setup that I’ve missed?
>

Hi Matteo,

When power management is not available, we cannot do failover to other host.

In 4.1 you can use a VM lease for your HA VM, enabling failover without
power
management.

Note that the lease cannot be on Ceph storage domain, you will have to add
other type of storage for the leases, since we don't support leases on ceph
yet.

Please check the vm-leases feature page for more info:
https://www.ovirt.org/develop/release-management/features/storage/vm-leases/

Nir


>
> Thank you
> Regards
> Matteo
>
> ___
> Users mailing list
> Users@ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

[ovirt-users] VMs HA with cinder volumes

[Matteo Dacrema]

Hi all,

I’ve a setup that uses Ceph through ovirt for VM storage.
I tested out VM HA and all works well when IPMI is reachable.
When it’s not reachable because of network or power failure VM states switch to 
unknown and it’s not restarted to other nodes.

Is it a specific setup that I’ve missed?

Thank you
Regards
Matteo

___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users