[Gluster-users] Enterprise Application with O_DIRECT access
Hideo Umemura(Max-T)
h.umemura at max-t.jp
Wed Mar 11 06:17:33 UTC 2009
Hi Amar-san,
Thank you for quick reply.
Is "glusterfs patched fuse" as you said fuse-2.7.3glfs10?
I tested glusterfs-2.0.0rc2 and fuse-2.7.3glfs10.
Or, is the patch for fuse kernel module made myself?
Regards,
hideo
Amar Tumballi wrote:
> With glusterfs patched fuse you should be able to work fine with
> O_DIRECT flag. The default stock kernel module doesn't support it.
>
> Yes, by increasing the blocksize of the filesystem, few application
> performance will increase significantly.
>
> Regards,
> Amar
>
> On Tue, Mar 10, 2009 at 10:52 PM, Hideo Umemura(Max-T)
> <h.umemura at max-t.jp <mailto:h.umemura at max-t.jp>> wrote:
>
> Hello everyone,
>
> I am learning and evaluating a glusterfs for film/video editing
> facilities.
> Some major film/video editing realtime applications are using the
> O_DIRECT file access for video/audio data files.
> The GLFS client via fuse mechanism is disallow the open file with
> O_DIRECT flag.
> I made a little sample code for read a file with O_DIRECT flag, and
> tried open the files on GLFS volumes.
> It returned "invalid arg" when open the file. It is OK when open the
> file without O_DIRECT flag.
>
> Is there a reason to have to refuse Direct I/O by the application
> though
> it seems to use Direct I/O internally of GLFS?
>
> Does someone have a problem same as me?
> Let me know if you have any advise,plan in future or patch on this.
>
> And, regarding the block size in fuse, I tried the optimization the
> block size in client side for film/video applications.
> When it was increased, the performance results is good.
> I think that it is convenient if it can change the block size in mount
> options.
> What do you think?
>
> Regards,
>
> Hideo.
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org <mailto:Gluster-users at gluster.org>
> http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users
>
>
>
>
> --
> Amar Tumballi
>
More information about the Gluster-users
mailing list