Please summarize state of ext4 resize >16TB

Please summarize state of ext4 resize >16TB

[Ryan, Michael]

I've been searching for info on the ability to resize ext4 from <16TB to >16TB.  This was not supported in the past and I've received conflicting information as to if/when this is/will be supported.  I've also seen indications that on disk data structures must be changed for this resize to work.

Is this capability currently supported?  
If so, what kernel level (and resize2fs level) is necessary?  Are there restrictions as to options at mkfs time?
If not, is this planned for a future release?  Or is this just not possible or practical due to on disk structures and is likely to never be supported?

Thank you.

Re: Please summarize state of ext4 resize >16TB

[Darrick J. Wong]

On Wed, Jan 08, 2014 at 12:14:17PM -0700, Ryan, Michael wrote:
> I've been searching for info on the ability to resize ext4 from <16TB to
> >16TB.  This was not supported in the past and I've received conflicting
> >information as to if/when this is/will be supported.  I've also seen
> >indications that on disk data structures must be changed for this resize to
> >work.
> 
> Is this capability currently supported?  
> If so, what kernel level (and resize2fs level) is necessary?  Are there
> restrictions as to options at mkfs time?  If not, is this planned for a
> future release?  Or is this just not possible or practical due to on disk
> structures and is likely to never be supported?

In most cases it's necessary to turn on 64bit (the feature flag) to cross the
16T boundary.  I've had patches against resize2fs to add this ability--it
involves moving data blocks and rewriting group descriptors--on the list for a
few months, but I don't know of Ted's plans to put them in a release.  Probably
post-1.43, is my guess.

I don't have any kernel patches to turn on 64bit, and I haven't any current
plans to make any.

--D
> 
> Thank you.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Please summarize state of ext4 resize >16TB

[Theodore Ts'o]

On Wed, Jan 08, 2014 at 11:30:18AM -0800, Darrick J. Wong wrote:
> 
> In most cases it's necessary to turn on 64bit (the feature flag) to cross the
> 16T boundary.  I've had patches against resize2fs to add this ability--it
> involves moving data blocks and rewriting group descriptors--on the list for a
> few months, but I don't know of Ted's plans to put them in a release.  Probably
> post-1.43, is my guess.

It's something that should eventually hit the master/next branch, so
it should be in an eventual 1.43 release.  Of course, I don't
currently have a timeline for when a 1.43 release would come out.  :-)

Probably in a few months if I had to guess, though.  The 1.42.x
release is pretty much locked down for all but the most trivial new
features, so I will be shortly switching my e2fsprogs focus to testing
and merging in the new features for the 1.43 release.

> I don't have any kernel patches to turn on 64bit, and I haven't any current
> plans to make any.

Turning on 64-bit requires rewriting the block group descriptors, and
searching the inodes to relocate data blocks to make room for the
expanded block group descriptors.  As such, I don't expect that we
would ever support on-line enablement of the 64-bit feature.  If
you've created a file system without the 64-bit feature, or have a
legacy file system that was created previously, you will have to
unmount the file system in order to set the 64-bit feature.

Regards,

						- Ted