Confused about samba4 & s3fs

[steve]

On 16/08/12 10:51, Rowland Penny wrote:
> Hi, over on the samba-users forum, somebody asked a question about
> Samba4's rfc2307 compatibility with Samba3 and got this reply:

>
>  From these two statements from '[ANNOUNCE] Samba 4.0 beta6', my
> understanding is that 's3fs' can & should be used to test it, is this
> correct? and if not, why not.
>

Hi Rowland, hi everyone

I think the intention is that we test s3fs and report errors against it.

I have one bug open on it concerning acls and file ownership on g+s shares:
  https://bugzilla.samba.org/show_bug.cgi?id=8938
ACL's in general seem to report strange values when viewed on an ext4 
filesystem. But that's about it. The acl's report differntly to how they 
behave but GPO's under sysvol with XP and 7 work fine for us under s3fs.

In my tests, s3fs seems stable and robust and I was surprised that I was 
told that it was not the recommended way to serve files. If I don't 
serve files with it then how can I test it?

I have since installed a S3.6 Vbox machine to do the file serving and 
can only say that it really is a 'pita' to have to do so. Yeah, OK it 
works but it's a lot of hassle and time spent on something which for us 
s3fs on the DC works almost perfectly.

It certainly is not clear what the official Samba stance is on this.

Are we saying that the DC and file server should be separate (in which 
case how about a S4 DC and a S4 s3fs file server on another box) or are 
we saying that all file serving _must_ be done from a S3 box which is 
separate from the DC?

My fear is that during development, s3fs may become unstable and we 
shall be forced to use S3 smbd on another box method.

Could anyone give us a definitive yes/no to s3fs?

Cheers,
Steve