On 12/21/14 06:59, Nir Soffer wrote:
----- Original Message -----
> From: "Julian De Marchi" <julian(a)jdcomputers.com.au>
> To: "Vered Volansky" <vered(a)redhat.com>
> Cc: users(a)ovirt.org
> Sent: Wednesday, December 17, 2014 4:29:36 PM
> Subject: Re: [ovirt-users] [STORAGE] Adding posix compliant FS
>
> On 17/12/14 21:38, Vered Volansky wrote:
>> Hi Julian,
>
> Hey Vered,
>
>> IIUC the you have a self-created lvm, to which you're trying to connect as
>> posix using ovirt.
>> If that's the case please try to manually change the device's
permissions
>> to 36:36 under /dev/<your_VG>.
>> If that doesn't work try the actual device it's mapped to (which is what
>> vdsm actually tries to mount).
>> If that also doesn't work please send us the output of:
>>>> ls -lh /dev/<your_VG> .
>
> Still failed mate after a chmod 36:36. The thing is though, it _does_
> mount the device for a moment before failing and unmounting it...
>
> [root@blade01 ~]# ls /dev/VolGroup/lv_images -lh
> lrwxrwxrwx 1 root root 7 Dec 15 16:52 /dev/VolGroup/lv_images -> ../dm-2
> [root@blade01 ~]# ls /dev/dm-2 -lh
> brw-rw---- 1 vdsm kvm 253, 2 Dec 15 16:52 /dev/dm-2
Hi Julian,
Hey Nir,
It is not clear what are you trying to do - is this lv on shared
storage, and accessed
by multiple hosts or use only one host?
It's a local LVM mount.
If the lv is shared, you cannot create a file system and mount it on
multiple hosts (unless you
are using cluster file system).
If this local storage for one host, you should create a file system and mount it, and
then
you can use the mountpoint as local storage.
If I proceed this way then the ovirt node needs to be in it's own local
cluster, correct?
Please describe in details what are you trying to do and how did you
create the storage
domain in the engine side.
I first did a default centos install which created the LVM stuff. I then
lvrename the home LV to images LV.
My end goal is to use the posix compliant mount option to have each of
my ovirt nodes the option to use local storage for VMs for which I'm not
fussed if they can migrate between the cluster or not.
Does this make more sense?
--julian