Linux-Misc Digest #259, Volume #26 Tue, 7 Nov 00 18:13:03 EST
Contents:
Re: smb vs nfs (Manfred Bartz)
Re: signal 11 and others while compiling (Vilmos Soti)
Re: Why, ext2 don't need defrag (Johan Kullstam)
Re: Secure mail under Red Hat 6.2 ("Matt Vanderveer")
Re: HELP!! Partition Magic Error 105. PartitionInfo inside: Please look!! ("Nataraj
Dasgupta")
Re: konqueror browser won't browse. (Jonathan)
Re: scsi module not loading at boot, but can insmod it ("starbux")
Re: Regular Expression Subsitution (NF Stevens)
Re: An appeal to experts for help! ("starbux")
Re: What happens to the Linux system when there is little or no HD space left
(Robert Heller)
Re: smb vs nfs ("starbux")
Re: Why, ext2 don't need defrag ("starbux")
rpm catch-22 ("Gerald Pollack")
Re: Why, ext2 don't need defrag (Robert Heller)
----------------------------------------------------------------------------
Crossposted-To: comp.os.linux.help
Subject: Re: smb vs nfs
From: Manfred Bartz <[EMAIL PROTECTED]>
Date: Tue, 07 Nov 2000 21:22:57 GMT
Victor Rajewski <[EMAIL PROTECTED]> writes:
> I have a small LAN at home (3 computers) in which I want to set up
> my fast(ish) linux box as a fileserver for the other two to run
> linux off (basically I want to export the /usr and /home dirs). in
> terms of speed and reliability, is smb or nfs a better option?
> security is not much of an issue as I trust my housemates (and they
> have physical access to the computer anyway).
SMB if faster and more secure than NFS, but it doesn't understand
permissions and ownership nor does it distinguish case. SMB's lack of
permissions and ownership handling may turn out to be a major nuisance
for you.
On a physically secure LAN and between hosts which run some sort of
UNIX, NFS is IMO the better choice.
--
Manfred
------------------------------
Subject: Re: signal 11 and others while compiling
From: Vilmos Soti <[EMAIL PROTECTED]>
Date: Tue, 07 Nov 2000 21:40:51 GMT
Jong-Min Park <[EMAIL PROTECTED]> writes:
> Whenever doing some heavy compilation (kernel, etc.) I get the
> following errors randomly
>
> signal 11 fault
http://www.bitwizard.nl/sig11
Vilmos
------------------------------
Subject: Re: Why, ext2 don't need defrag
From: Johan Kullstam <[EMAIL PROTECTED]>
Date: 07 Nov 2000 16:49:57 -0500
Robert Heller <[EMAIL PROTECTED]> writes:
> [EMAIL PROTECTED] (Christopher Browne),
> In a message on Sun, 05 Nov 2000 06:21:15 GMT, wrote :
>
> CB> In our last episode (Sat, 04 Nov 2000 21:13:34 -0500),
> CB> the artist formerly known as dan said:
> CB> >I understand that the ext2 filesystem is a little "smarter" then the fat
> CB> >fs, and it does not need to be defrag. But can someone explain why, I
> CB> >mean the physical architecture of how the ext2 fs works, or if it's too
> CB> >much to explain does anyone know of a site that can thoroughtly
> CB> >breakdown how the ext2 fs works.
> CB>
> CB> See:
> CB> <http://step.polymtl.ca/~ldd/ext2fs/ext2fs_toc.html>
> CB> Analysis of the Ext2fs structure
> CB>
> CB> As well as the references by Rémy Card and Theodore Ts'o that are
> CB> referenced therein.
> CB>
> CB> It is _NOT_ a "comparative analysis of ext2fs _as compared to DOS FAT_"
> CB> and thus will not provide a detailed answer as to _why_ ext2 is better.
> CB>
> CB> That is left as an exercise to the gentle reader; if you're not up
> CB> to looking at the sources and assessing it yourself, you would likely
> CB> not be able to find actual value in anything more specific than the
> CB> rather blank claim that "ext2 allocates files more intelligently than
> CB> FAT."
>
> Right. What would be more helpful would be an explaination as to why
> the FAT file system (MS-DOS/MS-Windows file system) both does fragment
> (which is not in *inself* bad) and why a fragmented FAT file system is
> bad, *partitularly* in the context of MS-DOS & MS-Windows.
>
> The simple / short explaination has to do with the fact that the FAT
> file system uses a *simple* linked list. Linked lists are notorious
> for fragmentation -- which is a major problem with LISP systems under
> memory-starved conditions -- they tend to page fault all over the place
> after things are running for a while.
in defense of lisp, lisp systems tend to have generational garbage
collection (GGC) to avoid this problem. you could view GGC as a kind
of defragmenter for memory. C and C++ don't allow the system to move
allocated data as they expose the raw pointers and the system would
have no way to fix them. C++ can easily be written in a malloc happy
style and is therefore very succeptible to the memory fragmentation
problem. this is why C++ speed tends to degrade and bog down after a
while whilst lisp keeps plodding ahead at a steady pace.
> This is *exactly* what happens
> with the FAT file system. Note: fragmentation itself is not really
> bad, if the operating system is designed to cope with the fragmentation
> *intellegently*, which usually means things like smart cacheing and
> various sorts of file system 'look ahead' / 'read ahead' features,
> which MS-DOS & MS-Windows generally lack (MS-DOS lacks these features,
> as does MS-Windows 95, 98, and ME -- 'NT and 2K are a little better).
> Note ext2 does fragment some, but both the ext2 structure and Linux
> itself are *designed* to cope with the fragmentation in an intellegent
> fashion and not suffer performance degradation because of it.
>
> The FAT file system was originally invented for *floppy disks*, which
> are relatively low capacity and where there is not really massive file
> creation and deletion. Also floppies are slow (in a relative sense) and
> don't have the spare sectors for a more complex file system. (Linux
> people tend to use the FAT filesystem for floppies, since Ext2 has too
> much overhead for floppies).
well, you have two cases
1) you need to move data to a dos/windows box and microsoft only does
FAT floppies.
2) you are storing for linux or other unix in which case any
filesystem at all is overkill for such a tiny medium. i use tar but
other prefer cpio.
thus it's no wonder ext2 floppies are rare.
> Fragmentation on a floppy is easy to cure:
> copy everything off the floppy, reformat it and copy everything
> back.
exactly. tar avoids the fragmentation issue. ;-)
> Actually floppies tend to wear out before fragmentation becomes serious.
> Or else the files on a floppy are 'static' (they are backup
> copies).
yes, and tar is good that too.
--
J o h a n K u l l s t a m
[[EMAIL PROTECTED]]
sysengr
------------------------------
From: "Matt Vanderveer" <[EMAIL PROTECTED]>
Subject: Re: Secure mail under Red Hat 6.2
Date: Tue, 07 Nov 2000 16:55:51 +0500
Crossposted-To: comp.os.linux.admin
Check out Postfix (www.postfix.org). It's got all of the features you're
looking for and is a drop-in replacement for Sendmail. I used it at my
last job and we're getting ready to switch over to it at my current job
(so long M$ Exchange!). I can't say enough good things about it.
Matt
In article <7X0M5.83833$[EMAIL PROTECTED]>, "D & S"
<[EMAIL PROTECTED]> wrote:
> Hi,
>
> I am looking at moving my company from using sendmail under Red Hat 6.2
> to something more secure and that gives me some levels of spam control.
>
> I have been trying out qmail, but am having trouble getting it to
> deliver incoming mail - it seems to drop the body of the message and
> only shows me the "From" and "To" fields in the header. If someone has
> qmail working fine under Red Hat 6.2, could you please give me some tips
> and at least let me know what your /var/qmail/rc file looks like?
>
> If any of you have some better ideas for me to give us better security
> and spam control, I'd greatly appreciate the help. Thanks!
>
> - Doug -
>
>
------------------------------
From: "Nataraj Dasgupta" <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux,comp.os.linux.help,comp.os.linux.questions,comp.os.linux.setup,comp.os.linux.x
Subject: Re: HELP!! Partition Magic Error 105. PartitionInfo inside: Please look!!
Date: Tue, 7 Nov 2000 16:57:44 -0500
Forget Partition Magic. The reason you could not boot linux is that the hdax
number changes when you repartition your drive. Use the boot disk that you
originally created and boot with linux root = /dev/hdax where x = the new
partition number,.. then, once you are in check /etc/fstab and make sure
your / is mounted in the correct /dev/hdax, otherwise make the appropriate
changes. I suggest searching on some newsgroup or at deja.com with the error
message, you will find enough references to see you through this problem.
"Brando" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
>
> I'm trying to recover my partition. I had it split up my 6GB HD into 2.5GB
Windows98 and
> 3.5GB Linux (Mandrake 7.1). I just recently could not boot up into Linux.
I got this error
> during bootup:
> > Kernel panic: VFS: Unable to mount root fs on 03:05
>
> So, I was going to just remove the Linux partition and freshly install LM
7.2
> But, when I run Partition Magic Pro 6.0 in DOS, I get Error 105:
> > Error 105 Partition starts on wrong boundary
> >
> > Cause of this problem:
> > The hard-disk partition table contains erroneous values. PartitionMagic
expects
> > FAT, NTFS, and HPFS partitions to begin and end on the boundaries used
by
> > FDISK. If they do not, the disk may be partially corrupted. In this
circumstance, if
> > PartitionMagic were to make any modifications it might cause loss of
data.
> > Therefore, PartitionMagic refuses to recognize any of the hard disk·s
partitions.
>
> So what can I do to resize my Windows98 partition to the full size of the
HD without
> damaging the Win98 partition?
>
> Here's the PartitionInfo output:
> >General System Information:
> > Total Physical Memory (bytes): 58,159,104
> > Used Physical Memory: (bytes): 54,263,808
> > Maximum Page File Size: (bytes): 237,502,464
> > Current Page File Size: (bytes): 48,635,904
> >
> >
> >
>
>===========================================================================
================================
> >Disk Geometry Information for Disk 1: 776 Cylinders, 240 Heads, 63
Sectors/Track
> >Warning: Logical drive chain points to sector without partition table.
> >System PartSect # Boot BCyl Head Sect FS ECyl Head Sect
StartSect NumSects
>
>===========================================================================
================================
> > 0 0 80 1 0 1 84 38 15 63
15,120 560,448
> >Info: Partition didn't end on cylinder boundary.
> > ucEndHead expected to be 239, not 15.
> >NOTEBOOK 0 1 00 38 16 1 0B 375 239 63
575,568 5,109,552
> >Error #105: Partition didn't begin on head boundary.
> > ucBeginHead expected to be 0 or 1, not 16.
> > 0 2 00 376 0 1 05 448 239 63
5,685,120 1,103,760
> >
> >
> >
>
>===========================================================================
================================
> >Partition Information for Disk 1: 5,729.1 Megabytes
> >Volume PartType Status Size MB PartSect # StartSect
TotalSects UsedSects FreeSects
>
>===========================================================================
================================
> > Free Space Pri 7.4 None -- 63
15,057 0 15,057
> > Type 84 Pri,Boot 273.7 0 0 15,120
560,448 560,448 0
> >C:NOTEBOOK FAT32 Pri 2,494.9 0 1 575,568
5,109,552 5,109,552 0
> > Extended Pri 538.9 0 2 5,685,120
1,103,760 63 1,103,697
> > EPBR Log 0.0 None -- 5,685,120
63 63 0
> > Free Space Log 538.9 None -- 5,685,183
1,103,697 0 1,103,697
> > Free Space Pri 2,414.2 None -- 6,788,880
4,944,240 0 4,944,240
> >
> >
>
>===========================================================================
================================
> >Boot Record for drive C: (Drive: 1, Starting sector: 575,568, Type:
FAT32)
>
>===========================================================================
================================
> > 1. Jump: EB 58 90
> > 2. OEM Name: was here
> > 3. Bytes per Sector: 512
> > 4. Sectors per Cluster: 8
> > 5. Reserved Sectors: 32
> > 6. Number of FAT's: 2
> > 7. Reserved: 0x0000
> > 8. Reserved: 0x0000
> > 9. Media Descriptor: 0xF8
> >10. Sectors per FAT: 0
> >11. Sectors per Track: 63 (0x3F)
> >12. Number of Heads: 240 (0xF0)
> >13. Hidden Sectors: 575568 (0x8C850)
> >14. Big Total Sectors: 5109552 (0x4DF730)
> >15. Big Sectors per FAT: 10861
> >16. Extended Flags: 0x0000
> >17. FS Version: 0
> >18. First Cluster of Root: 2 (0x2)
> >19. FS Info Sector: 1
> >20. Backup Boot Sector: 6
> >21. Reserved: 000000000000000000000000
> >22. Drive ID: 0x80
> >23. Reserved for NT: 0x00
> >24. Extended Boot Sig: 0x29
> >25. Serial Number: 0x125416D4
> >26. Volume Name: NOTEBOOK
> >27. File System Type: FAT32
> >28. Boot Signature: 0xAA55
>
> Please help! Thanx
> Brandon
------------------------------
From: Jonathan <[EMAIL PROTECTED]>
Subject: Re: konqueror browser won't browse.
Date: Tue, 07 Nov 2000 22:00:34 GMT
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Larry Autry) wrote:
> Trying to browse with konqueror in KDE2.
>
> "Could not create view for text/html
> Check your installation."
>
> Oh? Just what do I check? Everything is greyed out in Browser mode.
>
> BTW, KDE2 made it the default browser.
>
> Thanks,
> Larry
>
> --
>
> Larry Autry
> [EMAIL PROTECTED]
>
A friend is also wrestling with KDE2 and said you need the development
packages.
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: "starbux" <[EMAIL PROTECTED]>
Subject: Re: scsi module not loading at boot, but can insmod it
Crossposted-To: comp.os.linux.admin,comp.os.linux.hardware,comp.os.linux.setup
Date: Tue, 07 Nov 2000 22:12:56 GMT
Hi
Mandrake 7 has a new file in /etc called "modules". add your card's
module in there...it should work.
let me know if this works for u.
Dans l'article <[EMAIL PROTECTED]>, "Dennis Leiterman"
<[EMAIL PROTECTED]> a écrit :
> Hi,
> Just upgraded from RedHat 6.2 to RedHat 7.0. Was running 6.2 ok with
> no
> problems. After I upgraded, my adaptec 2930 scsi controller is not seen
> at boot. If I log in as root and do insmod aic7xxx it finds it and it
> works fine. I checked and in the /etc/modules.conf the line is there.
> I also have a line for my etherned adapter and sound card that seems to
> work fine. I tried moving it down in the file but that didn't work
> either. Thanks in advance.
>
> -Dennis
>
>
------------------------------
From: [EMAIL PROTECTED] (NF Stevens)
Subject: Re: Regular Expression Subsitution
Date: Tue, 07 Nov 2000 22:29:19 GMT
[EMAIL PROTECTED] (Fester) wrote:
[snip]
>Specifically, I need to search through an HTML file, find all url
>references within tags, and make them all lowercase. I'm moving a website
>from an NT Server to a Linux server, and the case sensetivities are giving
>me hell.
>
Try the following
awk -F\" -v OFS=\" '{
for (i = 2; i <= NF; i++)
{
if (match ($(i-1), "=$") && match ($i, "\\.htm$")
$i=tolower($i);
}
print;
}' something.htm >something.new.htm
Norman
------------------------------
From: "starbux" <[EMAIL PROTECTED]>
Subject: Re: An appeal to experts for help!
Crossposted-To: comp.os.linux.powerpc,comp.os.linux.hardware
Date: Tue, 07 Nov 2000 22:31:18 GMT
Hi,
i'm running a small linux network with a winblows machine and a imac
connected to it.
i've setup samba and the win pc shares and prints fine. i've setup
netatalk and the imac shares alright..but cant print.
my question: does MacOS X beta allow a remote-printer configuration like
linux or BeOS?
thx
Dans l'article
<[EMAIL PROTECTED]>, "Jack C
Pinette" <[EMAIL PROTECTED]> a écrit :
>
>
> On Thu, 2 Nov 2000, Anil Trivedi wrote:
>
>> BTW, since OS X is such a departure from OS 9, do your OS 9 programs
>> run fine under OS X or have you encountered problems?
>>
>> -Anil
>>
> I haven't had any problems, with one caveat... I'm on a modem, and
> 'classic' programs don't recognize Mac OS X Beta's PPP connections.
> This
> is a temporary situation, though. Other than that, any program that
> doesn't try to access hardware directly is fine. For example, while
> Quake
> 3 will run, it doesn't have any hardware acceleration. Of course,
> there's
> a Mac OS X Beta version of Quake 3 now, so no big issue. :)
>
> There are lots of places where you can get more info on Mac OS X, like
> Apple's site, osxtalk.com, osx.macnn.com, macaddict.com, macintouch.com,
> macworld, macweek, etc. etc.
>
> -Jack
>
------------------------------
From: Robert Heller <[EMAIL PROTECTED]>
Subject: Re: What happens to the Linux system when there is little or no HD space left
Date: Tue, 07 Nov 2000 22:39:46 -0000
mike <[EMAIL PROTECTED]>,
In a message on Tue, 07 Nov 2000 20:37:32 GMT, wrote :
m> Hi,
m> I was wondering what happens if you are working on a Linux
m> system and get low on hard disk space memory or run out of space.
m> Does the system give you a warning message, do programs freeze or
m> crash, etc?
Depends. Generally, Linux won't crash, but if you have a pathological
process things can get hairy.
Normally, the Ext2 file system (which Linux uses as its native file
system), leaves 15% of the disk for root only -- this means that a user
process will get a 'disk full' error while there is still some disk
space available. System daemons will still work and won't crash, etc.
'root' can go in and attempt to deal with the full disk issues -- such
as gziping stuff (root has that 'spare' 15% to use to create temp files,
such as the compressed file gzip creates while compressing).
If (virtual) memory runs out, you will not be able to create new
processes. This can make it hard to 'kill' out-of-control processes
(since you won't be able to run the ps or kill commands). There is a
'trick' to create this situation, something called a 'fork bomb'. I
believe Linux and most modern UNIX systems are able to survive a fork
bomb, but it is possible to write a fork bomb type program that could
get past the various safeguards. Of course a fork bomb is
*pathological* by nature -- normal programms are not fork bombs (at
least they should not be).
Having said all of this, (and had problems which involved a kind of
slow-motion fork bomb -- I had problems with a script relating to some
FidoNet code and improper device locking), I have found Linux to be
very robust in the face of some really bad situations. I had a '486
with only 16 meg of RAM and managed to create a sort of 'traffic jam'
of processes stuck because of problems with a cron job and a locked
device -- the cron job kept creating a new process that wedged on the
locked device, a new one every 15 minutes. It did this overnight and
in the morning I found this 'mess'. So long as the processes were just
waiting the system was actually running fine. Just that ps was
spitting out lots of output and Netscape was pagefaulting more than
usual... So it did a 'killall' on the processes and the system
started pagefaulting and became really, really 'busy' -- basically it
effectly was 'hung' for about 20 minutes while the system cleared out
all of the (wedged) processes. Eventually, things cleared up and
everything returned to normal. No crash or any other problems. I have
since fixed the FidoNet cron job script to check the lock *before*
going after the serial port and if the lock is 'old', kills off the
'stuck' FidoNet job and generally cleans things up and so on.
m> Thanks
m> Mike
m>
m>
--
\/
Robert Heller ||InterNet: [EMAIL PROTECTED]
http://vis-www.cs.umass.edu/~heller || [EMAIL PROTECTED]
http://www.deepsoft.com /\FidoNet: 1:321/153
------------------------------
From: "starbux" <[EMAIL PROTECTED]>
Subject: Re: smb vs nfs
Crossposted-To: comp.os.linux.help
Date: Tue, 07 Nov 2000 22:41:12 GMT
Hi,
this isnt the problem really...unless i'm mistaken about your setup,
samba and nfs are two different things. samba allows file/printer sharing
between windows machines and linux machine(s).
nfs mounts network shares between linux machines only.
Dans l'article
<[EMAIL PROTECTED]>, "Victor
Rajewski" <[EMAIL PROTECTED]> a écrit :
> heyheyhey
>
> I have a small LAN at home (3 computers) in which I want to set up my
> fast(ish) linux box as a fileserver for the other two to run linux off
> (basically I want to export the /usr and /home dirs). in terms of speed
> and reliability, is smb or nfs a better option? security is not much of
> an issue as I trust my housemates (and they have physical access to the
> computer anyway).
>
> anything else worth considering?
>
> vik
>
> [EMAIL PROTECTED] http://www.progsoc.uts.edu.au/~vik PGP:
> http://www.progsoc.uts.edu.au/~vik/pgp.txt
>
>
------------------------------
From: "starbux" <[EMAIL PROTECTED]>
Subject: Re: Why, ext2 don't need defrag
Crossposted-To: comp.os.linux.system
Date: Tue, 07 Nov 2000 22:49:12 GMT
Hi,
SGI's IRIX has that "journalized" feature...it needs no defrag.
there is a "defrag" utility for the linux fs, ext2. never used it tho...
i read somewhere that linux next fs, ext3, will have that "auto-defrag"
feature.
unless...my memory is bad ;-)
Dans l'article <5DJN5.59540$[EMAIL PROTECTED]>,
[EMAIL PROTECTED] a écrit :
> Robert Heller <[EMAIL PROTECTED]> writes:
>> [EMAIL PROTECTED] (Paul Lew), In a message on Sun, 05 Nov 2000 17:36:15
>> GMT, wrote :
>> PL> ??? SuSE 6.3 has the following msg: PL> EXT2-fs warning:
>> checktime reached, running e2fsck is recommended PL> PL> Doesn't
>> e2fsck also "defrag" a bit as well as checking the file-system?
>>
>> No. It just checks the file system. The "EXT2-fs warning: checktime
>> reached, running e2fsck is recommended" is not specific to SuSE. It is
>> a normal part of the EXT2-fs code.
>
> That's not 100% fair; supposing e2fsck finds problems, it may wind up
> shifting the broken files around.
>
> Mind you, I'm not sure that I'd count the ability to reorganize
> _broken_ files to be a major benefit :-). (Thus, it's only less than
> 100% fair by a Very Slim Margin...)
------------------------------
From: "Gerald Pollack" <[EMAIL PROTECTED]>
Subject: rpm catch-22
Date: Tue, 07 Nov 2000 23:01:11 GMT
I have a Mandrake 7.0 system. I'm attempting to upgrade printtool; I
currently have printtool-3.44-mdk. I've downloaded
printtool-3.55-1.noarch.rpm from rpmfind. When I give the command rpm -U
printtool-3.55-1.noarch.rpm I get:
only packages with major numbers <= 3 are supported by this version of RPM
error: printtool-3.55-1.noarch.rpm cannot be installed
My current version of rpm is 3.03-43mdk. The most recent version of rpm
listed on rpmfind is 4.04, but of course I get the same message when
trying to upgrade to it! Any suggestions?
Thanks,
--
G. Pollack Dept. of Biology, McGill Univ.
------------------------------
From: Robert Heller <[EMAIL PROTECTED]>
Subject: Re: Why, ext2 don't need defrag
Date: Tue, 07 Nov 2000 23:02:23 -0000
Johan Kullstam <[EMAIL PROTECTED]>,
In a message on 07 Nov 2000 16:49:57 -0500, wrote :
JK> Robert Heller <[EMAIL PROTECTED]> writes:
JK>
JK> > [EMAIL PROTECTED] (Christopher Browne),
JK> > In a message on Sun, 05 Nov 2000 06:21:15 GMT, wrote :
JK> >
JK> > CB> In our last episode (Sat, 04 Nov 2000 21:13:34 -0500),
JK> > CB> the artist formerly known as dan said:
JK> > CB> >I understand that the ext2 filesystem is a little "smarter" then the fat
JK> > CB> >fs, and it does not need to be defrag. But can someone explain why, I
JK> > CB> >mean the physical architecture of how the ext2 fs works, or if it's too
JK> > CB> >much to explain does anyone know of a site that can thoroughtly
JK> > CB> >breakdown how the ext2 fs works.
JK> > CB>
JK> > CB> See:
JK> > CB> <http://step.polymtl.ca/~ldd/ext2fs/ext2fs_toc.html>
JK> > CB> Analysis of the Ext2fs structure
JK> > CB>
JK> > CB> As well as the references by Rémy Card and Theodore Ts'o that are
JK> > CB> referenced therein.
JK> > CB>
JK> > CB> It is _NOT_ a "comparative analysis of ext2fs _as compared to DOS FAT_"
JK> > CB> and thus will not provide a detailed answer as to _why_ ext2 is better.
JK> > CB>
JK> > CB> That is left as an exercise to the gentle reader; if you're not up
JK> > CB> to looking at the sources and assessing it yourself, you would likely
JK> > CB> not be able to find actual value in anything more specific than the
JK> > CB> rather blank claim that "ext2 allocates files more intelligently than
JK> > CB> FAT."
JK> >
JK> > Right. What would be more helpful would be an explaination as to why
JK> > the FAT file system (MS-DOS/MS-Windows file system) both does fragment
JK> > (which is not in *inself* bad) and why a fragmented FAT file system is
JK> > bad, *partitularly* in the context of MS-DOS & MS-Windows.
JK> >
JK> > The simple / short explaination has to do with the fact that the FAT
JK> > file system uses a *simple* linked list. Linked lists are notorious
JK> > for fragmentation -- which is a major problem with LISP systems under
JK> > memory-starved conditions -- they tend to page fault all over the place
JK> > after things are running for a while.
JK>
JK> in defense of lisp, lisp systems tend to have generational garbage
JK> collection (GGC) to avoid this problem. you could view GGC as a kind
JK> of defragmenter for memory. C and C++ don't allow the system to move
JK> allocated data as they expose the raw pointers and the system would
JK> have no way to fix them. C++ can easily be written in a malloc happy
JK> style and is therefore very succeptible to the memory fragmentation
JK> problem. this is why C++ speed tends to degrade and bog down after a
JK> while whilst lisp keeps plodding ahead at a steady pace.
Right. *Early* lisp systems were very succeptible to the memory
fragmentation -- *modern* lisp use the hemispherial GC's (this is
probably different name for the generational garbage collection scheme.
JK>
JK> > This is *exactly* what happens
JK> > with the FAT file system. Note: fragmentation itself is not really
JK> > bad, if the operating system is designed to cope with the fragmentation
JK> > *intellegently*, which usually means things like smart cacheing and
JK> > various sorts of file system 'look ahead' / 'read ahead' features,
JK> > which MS-DOS & MS-Windows generally lack (MS-DOS lacks these features,
JK> > as does MS-Windows 95, 98, and ME -- 'NT and 2K are a little better).
JK> > Note ext2 does fragment some, but both the ext2 structure and Linux
JK> > itself are *designed* to cope with the fragmentation in an intellegent
JK> > fashion and not suffer performance degradation because of it.
JK> >
JK> > The FAT file system was originally invented for *floppy disks*, which
JK> > are relatively low capacity and where there is not really massive file
JK> > creation and deletion. Also floppies are slow (in a relative sense) and
JK> > don't have the spare sectors for a more complex file system. (Linux
JK> > people tend to use the FAT filesystem for floppies, since Ext2 has too
JK> > much overhead for floppies).
JK>
JK> well, you have two cases
JK> 1) you need to move data to a dos/windows box and microsoft only does
JK> FAT floppies.
JK> 2) you are storing for linux or other unix in which case any
JK> filesystem at all is overkill for such a tiny medium. i use tar but
JK> other prefer cpio.
JK> thus it's no wonder ext2 floppies are rare.
JK>
JK> > Fragmentation on a floppy is easy to cure:
JK> > copy everything off the floppy, reformat it and copy everything
JK> > back.
JK>
JK> exactly. tar avoids the fragmentation issue. ;-)
JK>
JK> > Actually floppies tend to wear out before fragmentation becomes serious.
JK> > Or else the files on a floppy are 'static' (they are backup
JK> > copies).
JK>
JK> yes, and tar is good that too.
I tend to prefer to use a file system on the floppy -- it makes things
easier if one is moving a few random small files -- often these are
config files or such and come from various places and need to go to
various places -- I can 'cp' them one by one and not have to repeatedly
untar stuff or untar to a temp place. I no longer use floppies for
backups -- I use tapes and Zip carts (reformatted with Ext2 file
sysetms).
JK>
JK> --
JK> J o h a n K u l l s t a m
JK> [[EMAIL PROTECTED]]
JK> sysengr
JK>
--
\/
Robert Heller ||InterNet: [EMAIL PROTECTED]
http://vis-www.cs.umass.edu/~heller || [EMAIL PROTECTED]
http://www.deepsoft.com /\FidoNet: 1:321/153
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.misc) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Misc Digest
******************************
