* RE: Re: NFS as a Cluster File System.
@ 2003-01-14 16:01 Lever, Charles
2003-01-25 20:32 ` Re: NFS as a Cluster File System (Locking question) Tom McNeal
0 siblings, 1 reply; 2+ messages in thread
From: Lever, Charles @ 2003-01-14 16:01 UTC (permalink / raw)
To: 'Neil Brown', Alan Robertson; +Cc: Lorn Kay, nfs, linux-ha
> On Thursday January 9, alanr@unix.sh wrote:
> >
> > NFS V3 and before have problems with "cache coherency".
> That is, the
> > different nodes in the cluster are not guaranteed to see
> the same contents.
> >
> > I think this is supposed to be fixed in v4.
> >
>
> NFSv4 does not try to "fix" this. It makes no attempts at
> "cache coherency" beyond what NFSv2/3 provide which is "close
> to open" cohenrence, meaning that if only one process has a
> file open at a time, then everythnig will appear coherent,
> and if multiple processes have the file open at the same
> time, they need to use record locking.
well, coherency is partially addressed in NFSv4 with delegations.
a server can delegate a file to a client, allowing the client
to cache the file and trust that the server will notify it when
another client wants to access the file (read or write).
for an aggressively shared file, this doesn't perform well, but
NFS has always assumed that there is little concurrent sharing
of files.
this paradigm probably doesn't fit well with typical file
usage in clusters, where files are very very large, and many
nodes may be working on independent pieces of the same file
at the same time. in that case, record locking might be
best. however, on Linux, the client purges the entire file
from its cache when a file is locked, rather than just the
areas that were byte-range locked.
-------------------------------------------------------
This SF.NET email is sponsored by: FREE SSL Guide from Thawte
are you planning your Web Server Security? Click here to get a FREE
Thawte SSL guide and find the answers to all your SSL security issues.
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en
_______________________________________________
NFS maillist - NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: Re: NFS as a Cluster File System (Locking question)
2003-01-14 16:01 Re: NFS as a Cluster File System Lever, Charles
@ 2003-01-25 20:32 ` Tom McNeal
0 siblings, 0 replies; 2+ messages in thread
From: Tom McNeal @ 2003-01-25 20:32 UTC (permalink / raw)
To: Lever, Charles; +Cc: nfs
Hi -
> best. however, on Linux, the client purges the entire file
> from its cache when a file is locked, rather than just the
> areas that were byte-range locked.
The purge must occur at lock time (or lock attempt time), right? Or
does the metadata also keep lock status, and therefore should occur
even when just a stat is done? (I don't think this is the case, but I
thought I'd make sure, not having looked at it recently)
Tom
Lever, Charles wrote:
>>On Thursday January 9, alanr@unix.sh wrote:
>>
>>>NFS V3 and before have problems with "cache coherency".
>>>
>>That is, the
>>
>>>different nodes in the cluster are not guaranteed to see
>>>
>>the same contents.
>>
>>>I think this is supposed to be fixed in v4.
>>>
>>>
>>NFSv4 does not try to "fix" this. It makes no attempts at
>>"cache coherency" beyond what NFSv2/3 provide which is "close
>>to open" cohenrence, meaning that if only one process has a
>>file open at a time, then everythnig will appear coherent,
>>and if multiple processes have the file open at the same
>>time, they need to use record locking.
>>
>
> well, coherency is partially addressed in NFSv4 with delegations.
> a server can delegate a file to a client, allowing the client
> to cache the file and trust that the server will notify it when
> another client wants to access the file (read or write).
>
> for an aggressively shared file, this doesn't perform well, but
> NFS has always assumed that there is little concurrent sharing
> of files.
>
> this paradigm probably doesn't fit well with typical file
> usage in clusters, where files are very very large, and many
> nodes may be working on independent pieces of the same file
> at the same time. in that case, record locking might be
> best. however, on Linux, the client purges the entire file
> from its cache when a file is locked, rather than just the
> areas that were byte-range locked.
>
>
> -------------------------------------------------------
> This SF.NET email is sponsored by: FREE SSL Guide from Thawte
> are you planning your Web Server Security? Click here to get a FREE
> Thawte SSL guide and find the answers to all your SSL security issues.
> http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en
> _______________________________________________
> NFS maillist - NFS@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/nfs
>
>
--
Tom McNeal
(650)906-0761(cell)
(650)964-8459(fax)
Email: trmcneal@attbi.com
-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
NFS maillist - NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2003-01-25 20:31 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-01-14 16:01 Re: NFS as a Cluster File System Lever, Charles
2003-01-25 20:32 ` Re: NFS as a Cluster File System (Locking question) Tom McNeal
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.