[Gluster-users] using KVM on glusterfs

George Machitidze giomac at gmail.com
Mon Sep 24 08:57:47 UTC 2012


O_DIRECT is supported for fuse since kernel 3.4

Best regards,
George Machitidze

On Mon, Sep 24, 2012 at 11:50 AM, Joe Julian <joe at julianfamily.org> wrote:

>  1. is a deficiency of the libvirt test for safe filesystems. There is a
> way to override that, but I'm not sure off the top of my head what that
> was. You should be able to search the libvirt mailing lists for that error
> and find it though. I know I read it there a little while ago.
> 2. Should work with any current kernels and the "direct-io-mode=yes" mount
> option (using the fuse client).
> 3. That's correct. qemu-kvm sets o_direct for just about anything but
> type=raw,cache=none. That's (imho) a qemu bug. Any app that doesn't fall
> back if it can't do O_DIRECT should.
> Anyway, I hope that all helps.
> On 09/24/2012 12:39 AM, flintcq wrote:
> 1. kvm can run on glusterfs using option cache=writeback, but can not
> implement migration because of following error:
> Unable to migrate guset: Unsafe migration: Migration may lead to data
> corruption if disks use cache != none
> 2. then i change the option as cache=none, but i doesn't work with
> following error:
> could not open disk image ..... invalid argument
>  3.later, i found that fuse doesn't support cache=none(using O_DIRECT
> flag), and if using disks with option type=raw, then cache=none may
> workable, so i change the type of disks, but  still the same error
> mentioned above.
>  this is my most works.
> **
>  **
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20120924/ed3a3821/attachment.html>

More information about the Gluster-users mailing list