[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [nfsv4] FedFS Meeting Minutes, 10/22/2009
- To: "LeMahieu, Paul" <LeMahieu_Paul at emc.com>, "Nicolas Williams" <Nicolas.Williams at sun.com>
- Subject: Re: [nfsv4] FedFS Meeting Minutes, 10/22/2009
- From: "Everhart, Craig" <Craig.Everhart at netapp.com>
- Date: Thu, 22 Oct 2009 17:51:29 -0400
- Cc: "Lentini, James" <James.Lentini at netapp.com>, nfsv4 at ietf.org
- Delivered-to: nfsv4 at core3.amsl.com
- In-reply-to: <ABB4EE6D-6B15-42D1-9895-CEB2D76AFCCE at emc.com>
- List-archive: <http://www.ietf.org/mail-archive/web/nfsv4>
- List-help: <mailto:nfsv4-request@ietf.org?subject=help>
- List-id: NFSv4 Working Group <nfsv4.ietf.org>
- List-post: <mailto:nfsv4@ietf.org>
- List-subscribe: <https://www.ietf.org/mailman/listinfo/nfsv4>, <mailto:nfsv4-request@ietf.org?subject=subscribe>
- List-unsubscribe: <https://www.ietf.org/mailman/listinfo/nfsv4>, <mailto:nfsv4-request@ietf.org?subject=unsubscribe>
- References: <alpine.LFD.2.00.0910221439290.11932 at jlentini-linux.nane.netapp.com> <CA9FE61D-C743-46AE-8B8F-04CAB8DDCE16 at emc.com> <E7372E66F45B51429E249BF556CEFFBC08A6C04F at RTPMVEXC1-PRD.hq.netapp.com> <20091022190822.GU892 at Sun.COM> <ABB4EE6D-6B15-42D1-9895-CEB2D76AFCCE at emc.com>
- Thread-index: AcpTYMezaWzVyBkhSRyfsqIzxHTsdQAAEYMg
- Thread-topic: [nfsv4] FedFS Meeting Minutes, 10/22/2009
In-line.
> -----Original Message-----
> From: LeMahieu, Paul [mailto:LeMahieu_Paul at emc.com]
> Sent: Thursday, October 22, 2009 5:44 PM
> To: Nicolas Williams
> Cc: Everhart, Craig; Lentini, James; nfsv4 at ietf.org
> Subject: Re: [nfsv4] FedFS Meeting Minutes, 10/22/2009
>
> See my comments inline [psl].
>
> > On Thu, Oct 22, 2009 at 02:59:42PM -0400, Everhart, Craig wrote:
> >> On Thursday, October 22, 2009 2:53 PM, LeMahieu, Paul wrote:
> >>> These topics have been discussed in the past. I'm trying
> to refresh
> >>> my memory on where we left things. Two questions:
> >>>
> >>> 1) Did we ever work on a doc describing the root
> fileset details
> >>> (database schema, etc)?
> >>> 2) How do file servers exporting the root fileset
> export both the
> >>> unified namespace root fileset, and other exports?
> >>> Are we expecting a separate IP for a server to export "/" of the
> >>> root fileset, so the physical path of other exports don't
> appear in
> >>> the namespace?
> >>
> >> (1) I don't think we ever did this.
> >
> > I'm not sure I understood the question. Was the question "how to
> > identify the root FSL for some domain?"? If so, a) I've asked this
> > before, b) I think one answer is: that's a local problem for root
> > servers. Not a very satisfying answer, that.
> >
> > It'd be nice to be able to store at least [and probably only] the
> > "root junctions" in a database, possibly the NSDB itself,
> or perhaps a
> > DNS TXT RR (you'll really want DNSSEC though!), so that root server
> > roles are trivial to configure (add the server location to the root
> > FSN's FSL list, twiddle a bit on the server.
> >
>
> [psl]
> My question was not about identifying the root FSL, but
> rather about the storing the entire "top-of-tree" root
> fileset, a collection of many top-level junctions, in the
> database. This makes it easier to deploy and manage root
> fileservers. The simplest use case I have is the /home use
> case. Today, I update an automount map, and my change is
> distributed. How will I add a new home user directory with Fed FS?
> Will I be able to update a database entry adding /home/bob,
> or will I have to update that on a physical file server, and
> rely on that filesystem being replicated by some proprietary
> means to other root file servers?
>
> --Paul
>
As James mentioned, this work is stalled, not forestalled.
As to the "proprietary" mention, I always thought we'd get to an IETF
protocol to replicate file systems, following some basic namespace work
with FedFS. And aren't LDAP replication schemes proprietary, today?
I rather like the idea of updating a physical file system when I want to
edit data. Perhaps I'm old-fashioned.
I'll stop now.
Craig