Re: namespace cache

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Just a quick question. Will glusterfs work on file systems that don't support inodes? I don't know much about the Windows NTFS details, but will glusterfs eventually work there or only targeted to work on the unix file systems? A few people have told me that they would use glusterfs to utilize free disk space on the windows office desktops as additional storage bricks. I'm using VMware player to currently create a unix virtual server within Windows to do this. Eventually I would like to have a simpler solution and have an agent work directly on the Win platform.


From: "Amar S. Tumballi" <amar@xxxxxxxxxxxxx>
To: "Brent A Nelson" <brent@xxxxxxxxxxxx>
CC: Gluster Developers List <gluster-devel@xxxxxxxxxx>
Subject: Re: namespace cache
Date: Wed, 4 Jul 2007 05:30:55 +0530

Let me tell how inode is propagated through each translator. Hope many may
want to know.

All the translators which has just one child nodes
(!(unify,afr,stripe,client-protocol,posix)) send the inode number passed by
the child volume to the upper layer. Hope that makes sense. Now, let me
explain each of these translators.

* posix - sends the inode number of the file/directory from actual file
system.

* client-protocol - sends inode number it got from the server to upper
layer.

* stripe - sends the inode number of the first child all the time (as stripe
needs first child node to be up all the time) to above layer.

* afr - sends the inode number of first available (in the order specified in
'subvolumes' option) child node to above layer.

* unify - always sends the inode number got from the namespace child.

So, if one uses the afr, it has a chance when a afr'ed namespace is down, it
gets different inode number from unify.

-bulde

On 7/4/07, Brent A Nelson <brent@xxxxxxxxxxxx> wrote:

Does this work even if the namespace volume is an AFR?

Thanks,

Brent


On Wed, 4 Jul 2007, Amar S. Tumballi wrote:

> Yes sir!
> The namespace cache gives persistant inodes accross mount/reboots. (If
> namespace data is not deleted in backend).
>
> -bulde
>
> On 7/4/07, Dale Dude <dale@xxxxxxxxxxxxxxx> wrote:
>>
>> Are you guys using the namespace cache for the inodes as well? Like
>> unionfs 2.0 ODF release does? Seems their method removes inode
>> collisions/loss and makes it easier for inode consistancy (inodes are
>> now permanent).
>>
>> Regards,
>> Dale
>>
>>
>> _______________________________________________
>> Gluster-devel mailing list
>> Gluster-devel@xxxxxxxxxx
>> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>>
>
>
>
> --
> Amar Tumballi
> http://amar.80x25.org
> [bulde on #gluster/irc.gnu.org]
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel@xxxxxxxxxx
> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>


_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel




--
Amar Tumballi
http://amar.80x25.org
[bulde on #gluster/irc.gnu.org]
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel

_________________________________________________________________
http://imagine-windowslive.com/hotmail/?locale=en-us&ocid=TXT_TAGHM_migration_HM_mini_2G_0507





[Index of Archives]     [Gluster Users]     [Ceph Users]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux