Re: [PATCH v6 6/9] nfsd: use the getattr operation to fetch i_version

[Dave Chinner <david@fromorbit.com>]

On Fri, Sep 30, 2022 at 02:34:51PM +0000, Chuck Lever III wrote:
> 
> 
> > On Sep 30, 2022, at 7:18 AM, Jeff Layton <jlayton@kernel.org> wrote:
> > 
> > Now that we can call into vfs_getattr to get the i_version field, use
> > that facility to fetch it instead of doing it in nfsd4_change_attribute.
> > 
> > Neil also pointed out recently that IS_I_VERSION directory operations
> > are always logged,
> 
> ^logged^synchronous maybe?

A pedantic note, but I think necessary because so many people still
get this wrong when it comes to filesystems and IO: synchronous !=
persistent.

Ext4 and XFS both use *asynchronous journalling* - they journal
changes first to memory buffers, and only make those recorded
changes persistent when they hit internal checkpoint thresholds or
something external requires persistence to be guaranteed.

->commit_metadata is the operation filesystems provide the NFS
server to *guarantee persistence*. This allows filesystems to use
asynchronous journalling for most operations, right up to the point
the NFS server requires a change to be persistent. "synchronous
operation" is a side effect of guaranteeing persistence on some
filesytems and storage media, whereas "synchronous operation"
does not provide any guarantee of persistence...

IOWs, please talk about persistence guarantees the NFS server
application requires and implements, not about the operations (or
the nature of the operations) that may be performed by the
underlying filesystems to provide that persistence guarantee.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com