[SUMM] Using tunefs

From: Susan Rodriguez <SUSROD_at_HBSI.COM>
Date: Thu, 13 Aug 1998 12:20:10 -0700

Thanks to all who replied. There is a lot of good info, which is
included below.

In summary, the man page on tunefs is wrong. You can't change the
minfree setting on a mounted filesystem. You CAN unmount the
filesystem, change the minfree, then remount. I tested it; it works
fine. If you do it mounted, it appears to work, but the available space
remains the same.

The optimization setting is determined by an algorithm, so the default
setting that you specify may not be used.

Thanks again to all,

susrod_at_hbsi.com
----------
From: Whitney Latta[SMTP:latta_at_decatl.alf.dec.com]
Sent: Tuesday, August 11, 1998 2:16 PM
To: Susan Rodriguez
Subject: Re: [Q] Using tunefs

Susan Rodriguez wrote:
>
> I am playing with the tunefs command on my test server. I want to
> reduce the minfree percentage from 10% to 5% as recommended in the man
> pages for filesystems of large capacity. On a few filesystems I may
> want to do 3%.
>
> I tried the tunefs for minfree of 5% on 2 filesystems (RAID groups
from
> a SWXCR card) each about 20GB. The filesystems showed no visible
change
> with df -k after tuning. tunefs does tell me that the change has been
> made, so why don't I see more space available?
>
> Also, tunefs recommends changing the optimization from "-o time" to
"-o
> space" if you reduce the minfree setting. Do I really need to do this
> for an extremely large filesystem? I am planning on making the
minfree


----------
From: Yizhong Zhou[SMTP:zhou_at_mathworks.com]
Sent: Tuesday, August 11, 1998 2:46 PM
To: Susan Rodriguez
Subject: Re: [Q] Using tunefs



You probably need to run tunefs on unmounted file systems to
the changes to take effect.


Yizhong


Yizhong Zhou zhou_at_mathworks.com, (508) 647-7371
The Mathworks, Inc. 24 Prime Park Way, Natick, MA
01760-1500


On Tue, 11 Aug 1998, Susan Rodriguez wrote:

>
> I am playing with the tunefs command on my test server. I want to
> reduce the minfree percentage from 10% to 5% as recommended in the man
> pages for filesystems of large capacity. On a few filesystems I may
> want to do 3%.
>
> I tried the tunefs for minfree of 5% on 2 filesystems (RAID groups
from
> a SWXCR card) each about 20GB. The filesystems showed no visible
change
> with df -k after tuning. tunefs does tell me that the change has been
> made, so why don't I see more space available?
>
> Also, tunefs recommends changing the optimization from "-o time" to
"-o
> space" if you reduce the minfree setting. Do I really need to do this
> for an extremely large filesystem? I am planning on making the
minfree
> change on some 35GB filesystems. I'd like to drop the minfree to 3%
> which would leave 900 MB still reserved. I would think that 900MB
would
> be plenty available space to leave the optimization for time.
>
> Any opinions or experiences?
>
> susrod_at_hbsi.com
>



> change on some 35GB filesystems. I'd like to drop the minfree to 3%
> which would leave 900 MB still reserved. I would think that 900MB
would
> be plenty available space to leave the optimization for time.
>
> Any opinions or experiences?
>
> susrod_at_hbsi.com


Woops, forgot your question about optimization.
Optimization automatically changes when the filesystem reaches a
predetermined percentage anyway, based in part on the minfree parameter.

With the size of your filesystems, I suspect they would be ok with the
default optimization (time). However if the filesystems habitually run
full or very near full, you may want to rethink this, depending on the
application mix.


Hope this helps.

-- 
=========================================
Whitney Latta
Compaq's Digital-Unix Support
Compaq Computer Services
1(800)354-9000
email: latta_at_alf.dec.com
=========================================
> 
> Any opinions or experiences?
> 
> susrod_at_hbsi.com
Tunefs(8) should be run on unmounted filesystems. If you run it on a
mounted filesystem, you may notice the changes will be zapped when the
next sync is called from update. You can check the changes using the
dumpfs(8) utility to look at the contents of the superblock:
	dumpfs /dev/<filesystem-raw-device> | grep -i minfree
If you've never seen the superblock contents, dumpfs(8) formats it into
something readable and you can see all the entries by simply piping it
into more(1) instead of the grep(1).
I hope this helps.
-- 
=========================================
Whitney Latta
Compaq's Digital-Unix Support
Compaq Computer Services
1(800)354-9000
email: latta_at_alf.dec.com
=========================================
----------
From: 	Dr. Tom Blinn, 603-884-0646[SMTP:tpb_at_doctor.zk3.dec.com]
Sent: 	Tuesday, August 11, 1998 2:00 PM
To: 	Susan Rodriguez
Subject: 	Re: [Q] Using tunefs 
> I am playing with the tunefs command on my test server.  I want to
> reduce the minfree percentage from 10% to 5% as recommended in the man
> pages for filesystems of large capacity.  On a few filesystems I may
> want to do 3%.
> 
> I tried the tunefs for minfree of 5% on 2 filesystems (RAID groups
from
> a SWXCR card) each about 20GB.  The filesystems showed no visible
change
> with df -k after tuning.  tunefs does tell me that the change has been
> made, so why don't I see more space available?
> 
> Also, tunefs recommends changing the optimization from "-o time" to
"-o
> space" if you reduce the minfree setting.  Do I really need to do this
> for an extremely large filesystem?  I am planning on making the
minfree
> change on some 35GB filesystems.  I'd like to drop the minfree to 3%
> which would leave 900 MB still reserved.  I would think that 900MB
would
> be plenty available space to leave the optimization for time.
> 
> Any opinions or experiences?
> 
> susrod_at_hbsi.com
Did you unmount and remount the file systems?  The change doesn't show
up
until you remount.
I believe that prior to V4.0D, you did have to make the change to -o
space
but in V4.0D and later, I believe it's ignored if there's a large disk
with
lots of real free space.
I have some UFS file systems where I run with 0% free, but they are
strictly
user data and I never put anything on them that matters if the file wind
up
truncated because I'm out of space.
Tom
 
 Dr. Thomas P. Blinn + UNIX Software Group + Compaq Computer Corporation
  110 Spit Brook Road, MS ZKO3-2/U20   Nashua, New Hampshire 03062-2698
   Technology Partnership Engineering           Phone:  (603) 884-0646
    Internet: tpb_at_zk3.dec.com           Digital's Easynet: alpha::tpb
     ACM Member: tpblinn_at_acm.org         PC_at_Home: tom_at_felines.mv.net
  Worry kills more people than work because more people worry than work.
      Keep your stick on the ice.        -- Steve Smith ("Red Green")
     My favorite palindrome is: Satan, oscillate my metallic sonatas.
                                         -- Phil Agre, pagre_at_ucsd.edu
     Yesterday it worked / Today it is not working / UNIX is like that 
			-- apologies to Margaret Segall 
  Opinions expressed herein are my own, and do not necessarily represent
  those of my employer or anyone else, living or dead, real or imagined.
 
> ----------
> From: 	alan_at_nabeth.cxo.dec.com[SMTP:alan_at_nabeth.cxo.dec.com]
> Sent: 	Tuesday, August 11, 1998 2:00 PM
> To: 	Susan Rodriguez
> Subject: 	Re: [Q] Using tunefs 
> 
> 
> 	If you made the change with the file system mounted it
> 	would produce the symtom you see.  And, when you do
> 	unmount the file system in the future, the change will
> 	probably disappear when the incore superblock gets
> 	written back.
> 
> 	From what I've seen, the file system itself will change
> 	time vs. space as it sees fit anyway.  I'm told there is
> 	an algorithm to it, but it doesn't behave that way in
> 	practice.
> 

Received on Thu Aug 13 1998 - 19:20:49 NZST