nfs-utils inode cache patch

nfs-utils inode cache patch

[Simon Peter]

[-- Attachment #1: Type: text/plain, Size: 317 bytes --]

Hi Neil,

okay, here's the patch making the kernel cache inode data of exported
directories forever, thus avoiding unnecessary disk spin-ups when this
data is looked for. It looks correct, works for me and both Bruce and
you have been encouraging it. :) Would be great if that could go into
nfs-utils.

Thanks,
Simon

[-- Attachment #2: nfs-utils-1.0.12-cache.dif --]
[-- Type: video/dv, Size: 493 bytes --]

[-- Attachment #3: Type: text/plain, Size: 345 bytes --]

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

[-- Attachment #4: Type: text/plain, Size: 140 bytes --]

_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: nfs-utils inode cache patch

[Neil Brown]

On Friday March 9, simon.peter@gmx.de wrote:
> Hi Neil,
> 
> okay, here's the patch making the kernel cache inode data of exported
> directories forever, thus avoiding unnecessary disk spin-ups when this
> data is looked for. It looks correct, works for me and both Bruce and
> you have been encouraging it. :) Would be great if that could go into
> nfs-utils.
> 
> Thanks,
> Simon

Hmmm...

> Content-Type: video/dv;
>  name="nfs-utils-1.0.12-cache.dif"

So it was a video/dv file was it?

Ok, I've applied it and added a big comment.

While reflecting I remembered that I like the fact that unused
information eventually disappeared from the kernel (where memory
cannot be pages out).  This information won't any more.

But that probably isn't a big cost.

http://neil.brown.name/git?p=nfs-utils;a=commitdiff;h=b1e1c8a05afdbbc3f217ef79c31c3b8ca302f9eb

Thanks,
NeilBrown


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: nfs-utils inode cache patch

[Talpey, Thomas]

At 12:22 AM 3/13/2007, Neil Brown wrote:
>http://neil.brown.name/git?p=nfs-utils;a=commitdiff;h=b1e1c8a05afdbbc3f217ef79c31c3b8ca302f9eb

Nit, typo: "log" -> "long"

* change underneath us, and an 'exportfs -f' can always
 * remove this from the kernel, so use a really log
 * timeout.  Maybe this should be configurable on the command


Tom.


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: nfs-utils inode cache patch

[J. Bruce Fields]

On Tue, Mar 13, 2007 at 03:22:11PM +1100, Neil Brown wrote:
> While reflecting I remembered that I like the fact that unused
> information eventually disappeared from the kernel (where memory
> cannot be pages out).  This information won't any more.
> 
> But that probably isn't a big cost.

The cost of the export structure itself should be very small.  I would
have guessed that the it's the mounts that are the bigger problem.  But
I think additional work is needed to allow us to keep only currently
used filesystems mounted.

And in any case if Olaf is right then it may not be the cost of keeping
those filesystems mounted so much as the time required to mount them all
at once on startup.  So that might be the problem to work on first.

But I wonder how much memory it takes to keep a few thousand mounts?
And in addition to the memory, are there any bits of VFS code that don't
scale well to lots of mounts?

--b.

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs