filesystems and "free space"

Thread View: comp.sys.sequent

4 messages

[T] Back to Threads [V] Tree View [G] Group Articles

4 total messages Started by Michael Faurot Fri, 02 Oct 1992 21:39

#437

Author: Michael Faurot
Date: Fri, 02 Oct 1992 21:39

32 lines
1963 bytes

Have the following hardware and OS:
	Balance B8 with Dynix 3.1.2

I've looked over the man pages for tunefs and newfs, and scanned
through the document about the BSD Fast File system in Vol. II of the
manuals, but I'm not real clear on what exactly is going on with this
free space being reserved.  According to the article about the FFS it
says something to the effect that having 10% free space makes things
work better and faster--how and why is this?  What happens when you
cut this down to about 5% or 2% or nothing at all? 

I've also noticed that when using newfs if I specify a value less than
10 to its -m option I get a warning message about "space optimazation"
is being used because less than 10% free.  What exactly does this mean?
What are the ramifications of initially creating a filesystem with
less than 10% free and this "space optimzation" in effect and then
later bumping it up to 10 or better via tunefs?  Does this in itself
cause the filesystem to stop using "space optimzation" and start using
whatever is more effecient?

Note, I get the general idea that having 10% makes things better, and
is the way one should normally create their filesystems, I
just want to understand why and what the situations are that arise
when it is less than 10% and for that matter more than 10%.

Thanks.
-- 

+--------------------+-------------------------------------------------------+
|   Michael Faurot   | Domain: mfaurot@bogart.UUCP		             |
|   ------- ------   | UUCP:   ...uunet!media!ka3ovk!irscscm!bogart!mfaurot  |
+--------------------+-------------------------------------------------------+

Re: filesystems and "free space"

#439

Author: B. Samuel Blanchard
Date: Mon, 05 Oct 1992 14:01

28 lines
1562 bytes

In article <1992Oct2.213912.1457@bogart.uucp> mfaurot@bogart.uucp (Michael Faurot) writes:
>Have the following hardware and OS:
>	Balance B8 with Dynix 3.1.2
>
>work better and faster--how and why is this?  What happens when you
>cut this down to about 5% or 2% or nothing at all? 
>|   Michael Faurot   | Domain: mfaurot@bogart.UUCP		             |

Quick snip: Motorola documentations states that 10% free may be excessive
in very large file systems.  It made sense to me.  My thought :  the
ability to locate the next "logical" block within a reasonable "physical"
location is "more greatly" influenced by available individual blocks than by
the total surface area.  I.e.  I can probably find a reasonable place for the
next block based on having 5% (99896 blocks) instead of 10% (199793) on my
preformatted 1gig disk.

responses encouraged.

The "space optimization" message interests me.  I have slightly smaller disks
on my S27 (792MB) and have always used std 10%.

Disclaimer: I think I understand the fast file system.  :-)

-- 
B. Sam Blanchard        sam@bsu-cs.bsu.edu
418 Winfield Dr         (317) 741-4500   work
Greenfield, IN 46140

Re: filesystems and "free space"

#440

Author: Kenneth Herron
Date: Mon, 05 Oct 1992 15:31

20 lines
1398 bytes

sam@bsu-cs.bsu.edu (B. Samuel Blanchard) writes:

>>work better and faster--how and why is this?  What happens when you
>>cut this down to about 5% or 2% or nothing at all? 

>Quick snip: Motorola documentations states that 10% free may be excessive
>in very large file systems.  
>I can probably find a reasonable place for the
>next block based on having 5% (99896 blocks) instead of 10% (199793) on my
>preformatted 1gig disk.

Sorry, no hard figures here, but we ran our news partition (350 megs or
so at the time) with 0% held back for quite a while.  Fragmentation
never became a problem, though of course we never allowed it to 
completely fill for any length of time.
-- 
Kenneth Herron                                   kherron@ms.uky.edu
University of Kentucky                              +1 606 257 2975
Department of Mathematics      I just thought of a neat disproof of
     Fermat's last theorem, but .signatures can only be four lines.

Re: filesystems and "free space"

#416

Author: Michael Faurot
Date: Thu, 08 Oct 1992 12:29

20 lines
1516 bytes

In article <1992Oct5.113145.28333@ms.uky.edu> kherron@ms.uky.edu (Kenneth Herron) writes:
>
>Sorry, no hard figures here, but we ran our news partition (350 megs or
>so at the time) with 0% held back for quite a while.  Fragmentation
>never became a problem, though of course we never allowed it to 
>completely fill for any length of time.

That's exactly the reasons I was asking about this in the first place.
I'm currently running a 30M partition which is exclusively for use by
news and nothing else.  I've used tunefs to change the free space from
the default 10% down to 2%.  

BTW: What do you do to determine the fragmentation?  Are you refering
to the messages generated by fsck, or some other utiltiy?
-- 

+--------------------+-------------------------------------------------------+
|   Michael Faurot   | Domain: mfaurot@bogart.UUCP		             |
|   ------- ------   | UUCP:   ...uunet!media!ka3ovk!irscscm!bogart!mfaurot  |
+--------------------+-------------------------------------------------------+

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads