From: Bryan Schumaker <bjschuma@netapp.com>
To: Sukwoo Kang <sukwoo.kang@gmail.com>
Cc: linux-nfs@vger.kernel.org
Subject: Re: NFSv4 read delegation
Date: Mon, 26 Sep 2011 08:52:24 -0400 [thread overview]
Message-ID: <4E807588.8040904@netapp.com> (raw)
In-Reply-To: <CAOB=RsJUwPsU4gjnW208wYJ-VCsbRovyg3xEfHn2boWwGicZ=w@mail.gmail.com>
On 09/23/2011 08:17 PM, Sukwoo Kang wrote:
> Hello, I'm trying to find out how NFSv4 client can get read
> delegation. I'm using separate RHEL6 machines (2.6.32-71.el6.x86_64)
> as server and client respectively.
I'm able to get a read delegation by holding two files open at the same time.
$ tail -f /nfs/file1.txt
$ tail -f /nfs/file2.txt
Will give me a read delegation in the second OPEN reply.
- Bryan
>
> If I use pynfs, I can see server gives out read delegation. Basically
> pynfs sends these RPCs to server - open (create), open_confirm, open,
> open_confirm.
>
> On real NFS client,
>
> If I do 'cat <file>' twice, client does not get read delegation.
> - (nfs4state.c)
> - if (!cb_up || !sop->so_confirmed)
> goto out;
> - Here sop->so_confirmed is 0.
>
> If I open the existing file with O_RDONLY twice (i.e., open, not close
> it and open again), client sends open only once to server.
> - So here also sop->so_confirmed is 0. Delegation is not sent out.
> (I've added more debug messages shown in below.)
>
> 2011-09-23T22:41:44.642938+02:00 int002st001 kernel: NFSD:
> nfs4_open_delegation enter
> 2011-09-23T22:41:44.642942+02:00 int002st001 kernel: NFSD: nfs4_open_claim_null
> 2011-09-23T22:41:44.643853+02:00 int002st001 kernel: NFSD: cb_up=1,
> sop->so_confirmed=0
> 2011-09-23T22:41:44.643857+02:00 int002st001 kernel: NFSD: either
> callback is not up or not open confirmed
> 2011-09-23T22:41:44.643858+02:00 int002st001 kernel: NFSD:
> nfs4_open_delegation exit
>
> If I open the existing file with O_RDWR and then open it with
> O_RDONLY, I can see "sop->so_confirmed = 1", but setlease failed.
> - I think it is because there is writer on this file.
>
> 2011-09-23T11:23:28.116280+02:00 int002st001 kernel: NFSD:
> nfs4_open_delegation enter
> 2011-09-23T11:23:28.116717+02:00 int002st001 kernel: NFSD: nfs4_open_claim_null
> 2011-09-23T11:23:28.117879+02:00 int002st001 kernel: NFSD: cb_up=1,
> sop->so_confirmed=1
> 2011-09-23T11:23:28.117887+02:00 int002st001 kernel: NFSD alloc_init_deleg
> 2011-09-23T11:23:28.117889+02:00 int002st001 kernel: NFSD: setlease
> failed [-11], no delegation
> 2011-09-23T11:23:28.119101+02:00 int002st001 kernel: NFSD:
> close_delegation dp ffff88006ba05e60
> 2011-09-23T11:23:28.119109+02:00 int002st001 kernel: NFSD: freeing dp
> ffff88006ba05e60
> 2011-09-23T11:23:28.120050+02:00 int002st001 kernel: NFSD:
> nfs4_open_delegation exit
>
> What is missing here? I will appreciate any advice. Thank you.
>
> Regards,
> Sukwoo Kang
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
prev parent reply other threads:[~2011-09-26 12:53 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-09-24 0:17 NFSv4 read delegation Sukwoo Kang
2011-09-26 12:52 ` Bryan Schumaker [this message]
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=4E807588.8040904@netapp.com \
--to=bjschuma@netapp.com \
--cc=linux-nfs@vger.kernel.org \
--cc=sukwoo.kang@gmail.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).