All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Mkrtchyan, Tigran" <tigran.mkrtchyan@desy.de>
To: Rick Macklem <rmacklem@uoguelph.ca>
Cc: linux-nfs <linux-nfs@vger.kernel.org>,
	trondmy@hammerspace.com,
	Anna Schumaker <Anna.Schumaker@Netapp.com>,
	Tom Haynes <loghyr@gmail.com>
Subject: Re: [PATCH] nfs4: flex_file: ignore synthetic uid/gid for tightly coupled DSes
Date: Mon, 20 Aug 2018 22:41:01 +0200 (CEST)	[thread overview]
Message-ID: <1195851744.36922369.1534797661849.JavaMail.zimbra@desy.de> (raw)
In-Reply-To: <YTOPR0101MB18201895B73ED2C4BA3B0840DD320@YTOPR0101MB1820.CANPRD01.PROD.OUTLOOK.COM>

Hi Rick,

draft-19 says:

   For tight coupling, ffds_stateid provides the stateid to be used by
   the client to access the file.  For loose coupling and a NFSv4
   storage device, the client will have to use an anonymous stateid to
   perform I/O on the storage device.  With no control protocol, the
   metadata server stateid can not be used to provide a global stateid
   model.  Thus the server MUST set the ffds_stateid to be the anonymous
   stateid.

To me, the last sentence sounds like a clear statement, that it's server
responsibility to provide either global or anonymous stateid and client
should just use it as is. IOW,

+	stateid = nfs4_ff_layout_select_ds_stateid(lseg, idx);
+	if (stateid)
+		nfs4_stateid_copy(&hdr->args.stateid, stateid);


must happen independent from ds being loose or tight coupled. As our DSes
use the same stateid's as MDS, we did see that provided stateid is not used.

Trond, Tom, can you comment on it?

Thanks,
    Tigran.

----- Original Message -----
> From: "Rick Macklem" <rmacklem@uoguelph.ca>
> To: "Tigran Mkrtchyan" <tigran.mkrtchyan@desy.de>, "linux-nfs" <linux-nfs@vger.kernel.org>
> Cc: trondmy@hammerspace.com, "Anna Schumaker" <Anna.Schumaker@Netapp.com>
> Sent: Monday, August 20, 2018 3:18:00 PM
> Subject: Re: [PATCH] nfs4: flex_file: ignore synthetic uid/gid for tightly coupled DSes

> I just put a patch for the stated part (stripped out my version of the cred
> stuff, which I noticed I missed commit in it;) so it might be easier to read.
> It is here:
> http://people.freebsd.org/~rmacklem/flexfilestateid.patch
> 
> Thanks for doing this, rick
> 
> ________________________________________
> From: linux-nfs-owner@vger.kernel.org <linux-nfs-owner@vger.kernel.org> on
> behalf of Rick Macklem <rmacklem@uoguelph.ca>
> Sent: Monday, August 20, 2018 8:51:14 AM
> To: Tigran Mkrtchyan; linux-nfs@vger.kernel.org
> Cc: trondmy@hammerspace.com; Anna.Schumaker@Netapp.com
> Subject: Re: [PATCH] nfs4: flex_file: ignore synthetic uid/gid for tightly
> coupled DSes
> 
> Thanks. I'll test this later to-day. (I did a slightly more complex version
> of the fix outside of the ff_layout_get_ds_cred() call, but yours is
> definitely simpler).
> 
> There is also the "stateid", which I believe is supposed to be the one in
> the layout for the tightly coupled case (for NFSv4 I/O ops to the DS).
> My patch is here and has the changes for stated as well as cred.
> 
> http://people.freebsd.org/~rmacklem/flexfile.patch
> (Sorry, I don't know git;-)
> 
> Thanks for doing this and I'll post if my testing finds problems, rick
> 
> 
> ________________________________________
> From: linux-nfs-owner@vger.kernel.org <linux-nfs-owner@vger.kernel.org> on
> behalf of Tigran Mkrtchyan <tigran.mkrtchyan@desy.de>
> Sent: Monday, August 20, 2018 2:56:08 AM
> To: linux-nfs@vger.kernel.org
> Cc: trondmy@hammerspace.com; Anna.Schumaker@Netapp.com; Tigran Mkrtchyan
> Subject: [PATCH] nfs4: flex_file: ignore synthetic uid/gid for tightly coupled
> DSes
> 
> for tightly coupled DSes client must ignore provided synthetic uid and
> gid as stated in draft-ietf-nfsv4-flex-files-19#section-5.1.
> 
> Signed-off-by: Tigran Mkrtchyan <tigran.mkrtchyan@desy.de>
> ---
> fs/nfs/flexfilelayout/flexfilelayoutdev.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> index d62279d3fc5d..290625cfd369 100644
> --- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> +++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> @@ -452,7 +452,7 @@ ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg, u32
> ds_idx,
>        struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
>        struct rpc_cred *cred;
> 
> -       if (mirror) {
> +       if (mirror && !mirror->mirror_ds->ds_versions[0].tightly_coupled) {
>                cred = ff_layout_get_mirror_cred(mirror, lseg->pls_range.iomode);
>                if (!cred)
>                        cred = get_rpccred(mdscred);
> --
> 2.17.1

  reply	other threads:[~2018-08-20 23:58 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-08-20  6:56 [PATCH] nfs4: flex_file: ignore synthetic uid/gid for tightly coupled DSes Tigran Mkrtchyan
2018-08-20 12:51 ` Rick Macklem
2018-08-20 13:18   ` Rick Macklem
2018-08-20 20:41     ` Mkrtchyan, Tigran [this message]
2018-08-20 20:52       ` Tom Haynes
2018-08-20 21:16         ` Rick Macklem
2018-08-20 21:28           ` Mkrtchyan, Tigran
2018-08-20 21:46             ` Rick Macklem
2018-08-21  1:32       ` Rick Macklem

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1195851744.36922369.1534797661849.JavaMail.zimbra@desy.de \
    --to=tigran.mkrtchyan@desy.de \
    --cc=Anna.Schumaker@Netapp.com \
    --cc=linux-nfs@vger.kernel.org \
    --cc=loghyr@gmail.com \
    --cc=rmacklem@uoguelph.ca \
    --cc=trondmy@hammerspace.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.