On 07/15/2013 06:38 PM, John Skinner wrote:
I dug a little bit deeper and found that it IS a supported storage type of NETFS, as a
valid poor format type (see below from libvirt.org). Now, not being familiar with how
CloudStack handles storage; I was think that under cloud-plugin-hypervisoer-kvm > src
> com.cloud.hypervisor.kvm.storage > LibvirtStorageAdaptor.java I create a pool
type similar to the one for NFS with the required information for using GlusterFS. Is
this assumption correct?
Yes, if that's the case, you could simply change "netfs" into
"glusterfs", but I'm not sure if that uses libgfapi, although that
wouldn't matter much.
It's when building the domain XML and actually invoking Qemu that you
want it to use libgfapi.
Wido
Thanks,
John
Valid volume format types
The valid volume types are the same as for the directory pool type.
Network filesystem pool
This is a variant of the filesystem pool. Instead of requiring a local block
device as the source, it requires the name of a host and path of an exported
directory. It will mount this network filesystem and manage files within the
directory of its mount point. It will default to using NFS as the protocol.
Example pool input
<pool type="netfs">
<name>virtimages</name>
<source>
<host name="nfs.example.com"/>
<dir path="/var/lib/virt/images"/>
</source>
<target>
<path>/var/lib/virt/images</path>
</target>
</pool>
Valid pool format types
The network filesystem pool supports the following formats:
auto - automatically determine format
nfs
glusterfs
cifs
On Jul 15, 2013, at 10:08 AM, John Skinner <[email protected]> wrote:
Wido,
Are you sure on that? I know the libgfapi is in C. But I thought GlusterFS was
now supported in both libvirt and qemu (1.0.1+, 1.3; respectively).
1.0.1: Dec 17 2012
Features:
Introduce virtlockd daemon (Daniel P. Berrange),
parallels: add disk and network device support (Dmitry Guryanov),
Add virDomainSendProcessSignal API (Daniel P. Berrange),
Introduce virDomainFSTrim() public API (Michal Privoznik),
add fuse support for libvirt lxc (Gao feng),
Add Gluster protocol as supported network disk backend (Harsh Prateek Bora),
various snapshot improvements (Peter Krempa, Eric Blake)
Thanks,
John
On Jul 15, 2013, at 9:56 AM, Wido den Hollander <[email protected]> wrote:
Hi John,
On 07/15/2013 04:52 PM, John Skinner wrote:
Thanks, Wido.
I am not a programmer per se, but I am going to pull the code down and have a
look to see if I can figure it out. I know some java guys so may be able to get
some help on that end.
It won't be only Java code, but also C code to manage the GlusterFS storage
pool in libvirt: http://libvirt.org/storage.html
Currently GlusterFS isn't supported in libvirt as a storage pool, but the
CloudStack agent relies on that.
It might be possible to do without libvirt, but I'm not sure how that would
work out.
Ceph and RBD are my thing, I'm not a GlusterFS expert.
Wido
Thanks,
John
On Jul 15, 2013, at 9:36 AM, Wido den Hollander <[email protected]> wrote:
Hi John,
On 07/15/2013 04:31 PM, John Skinner wrote:
Is there any way to use GlusterFS with the native QEMU libgfapi so we do not
have to use Fuse to access the shares? Or are there any plans to build libgfapi
QEMU support into CloudStack in the future?
As for now there is no way to use libgfapi with Qemu/KVM in CloudStack, nor are
there any plans to implement this.
Patches are welcome though! Would be great to see this be written.
Wido
Thanks,
John
