[prev in list] [next in list] [prev in thread] [next in thread]
List: ietf-nfsv4
Subject: [nfsv4] RE: blocks draft 01: operation issues
From: Black_David () emc ! com
Date: 2005-07-29 23:35:34
Message-ID: F222151D3323874393F83102D614E0550824CE () CORPUSMX20A ! corp ! emc ! com
[Download RAW message or body]
> I don't understand the following under operation issues:
>
> The LAYOUTCOMMIT operation appears to have some LAYOUTRETURN side
> effects. This should be separated so that LAYOUTCOMMIT does not
> return any layout extents; the required LAYOUTRETURN can be in a
> compound immediately after the LAYOUTCOMMIT.
>
> Questions:
> - What side effects?
>
> - How does LAYOUTCOMMIT return extents in the current proposal?
I thought I saw text in an older version that enabled extent return
via LAYOUTCOMMIT - whatever the text at issue was, it's not in the
current -03 version. My primary concern was server tracking of client
response to recalls - the tracking should only involve LAYOUTRETURN
ops. I'll go take that issue out of my working version of the pNFS
blocks draft.
> The way I view it, is that LAYOUTRETURN hands back the layout saying it
> is not going to be used anymore. Since LAYOUTCOMMIT is for handling
> writes, mtime should be only set at LAYOUTCOMMIT time. I could imagine
> atime being updated by LAYOUTRETURN, but only if it was a hint (similar
> to the way time is updated by LAYOUTCOMMIT). SETATTR should be used to
> explicitly set any times.
Seems ok. My primary concern was server tracking of client response
to a recall - the tracking should only involve LAYOUTRETURN ops.
Thanks,
--David
----------------------------------------------------
David L. Black, Senior Technologist
EMC Corporation, 176 South St., Hopkinton, MA 01748
+1 (508) 293-7953 FAX: +1 (508) 293-7786
black_david@emc.com Mobile: +1 (978) 394-7754
----------------------------------------------------
_______________________________________________
nfsv4 mailing list
nfsv4@ietf.org
https://www1.ietf.org/mailman/listinfo/nfsv4
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic