From: Neil Brown <neilb@suse.de>
Subject: Re: [PATCH 03/12] Extend the exportfs interface to
	pass	fslocations info into the kernel.
Date: Fri, 9 Feb 2007 11:39:45 +1100
Message-ID: <17867.49873.883604.237104@notabene.brown>
References: <20070208222606.23464.71348.stgit@rock.citi.umich.edu>
	<20070208222704.23464.86120.stgit@rock.citi.umich.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Cc: nfs@lists.sourceforge.net
To: Kevin Coffman <kwc@citi.umich.edu>
In-Reply-To: message from Kevin Coffman on Thursday February 8
Sender: nfs-bounces@lists.sourceforge.net
Errors-To: nfs-bounces@lists.sourceforge.net

On Thursday February 8, kwc@citi.umich.edu wrote:
> From: Kevin Coffman <kwc@citi.umich.edu>
> 
> Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
> Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
> 
> Extend exportfs interface to pass fslocations info into the kernel,
> using syntax modelled after AIX.  Adds "--refer" and "--replicas"
> options to exportfs to enable use of the kernel fslocation code.

I cannot see the "--refer" and "--replicas" implemented or documented
in the patch.  I think the changelog entry must be out of date, and
when we remove that sentence, there isn't much left....

> +/* Called by setting 'Method = stub' in config file.  Just returns
> + * some syntactically correct gibberish for testing purposes.
> + */
> +static struct servers *method_stub(char *key)
> +{
> +	struct servers *sp;
> +	struct mount_point *mp;
> +
> +	syslog(LOG_INFO, "called method_stub");
> +	sp = malloc(sizeof(struct servers));
> +	if (!sp)
> +		return NULL;
> +	mp = calloc(1, sizeof(struct mount_point));
> +	if (!mp) {
> +		free(sp);
> +		return NULL;
> +	}
> +	sp->h_num = 1;
> +	sp->h_mp[0] = mp;
> +	mp->h_host = strdup("stub_server");
> +	mp->h_path = strdup("/my/test/path");
> +	sp->h_referral = 1;
> +	return sp;
> +}


This makes perfect sense for testing during development, but do you
really want it in the released version?
If so, should there be "#ifdef DEBUG" or something around it ??


And I'm a little confused by the replicas_lookup, method_list stuff.
I think it is just parsing the format from /etc/exports so that it can
be written in the different format to the kernel.
It would seem more appropriate to do this when parsing the
/etc/exports (to /etab) file rather than every time we talk to the
kernel.  Is there are reason for the way it is?
(Just trying to make sure I understand).


Thanks,
NeilBrown

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier.
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs