Linux-Misc Digest #531, Volume #27                Wed, 4 Apr 01 17:13:02 EDT

Contents:
  Startup items in RH6.2 (Sizam)
  Re: I would like to register a complaint ... ([EMAIL PROTECTED])
  Re: what 'language' to learn ([EMAIL PROTECTED])
  Re: What are servers/services ? ([EMAIL PROTECTED])
  Re: The Python is so powerful, easy to use? compared with C++ (Elf Sternberg)
  Re: Startup items in RH6.2 (Lynne Lawrence)
  Re: I would like to register a complaint ... (hoffmyster)
  Dhcp ("Nelson Yip")
  Re: what 'language' to learn (Juergen Heinzl)
  Using RPM with RH5.2 (hoffmyster)
  Re: CDR writing with SCSI emulation stops ppp transfers ("Eric Potter")
  Re: CDR writing with SCSI emulation stops ppp transfers ("Eric Potter")
  Re: CDR writing with SCSI emulation stops ppp transfers ("Eric Potter")
  Re: Startup items in RH6.2 (Markku Kolkka)
  Re: I would like to register a complaint ... (Jeremiah DeWitt Weiner)
  Re: missing initab file.......enter runlevel? (Dranthony)

----------------------------------------------------------------------------

From: Sizam <[EMAIL PROTECTED]>
Subject: Startup items in RH6.2
Date: Wed, 04 Apr 2001 13:09:50 -0700

How do I find what programs get turned on at startup?  My server went 
down during a blackout and came back on but didn't startup sendmail and 
started up the wrong apache program, where can I change this?

Thanks,
Sam

------------------------------

From: [EMAIL PROTECTED]
Subject: Re: I would like to register a complaint ...
Date: Wed, 04 Apr 2001 20:15:51 GMT

John Thompson <[EMAIL PROTECTED]> writes:
> KCmaniac wrote:
> > I realize this isn't the argument clinic, but I would like to register a
> > complaint.
> > 
> > There is a very fundemental concept in the DOS/Windows world of being
> > able to format a partition, after which you can begin again compiling
> > data into that empty but very much functional partition.  WHY DOESN'T
> > LINUX HAVE AN EQUIVALENT COMMAND/FUNCTION???
> 
> Err... what's wrong with "mkfs?"

Hah, hah!  April Fools!

It is an April Fools joke, right?  Nobody could possibly be so stupid
as to think that...
-- 
(concatenate 'string "cbbrowne" "@ntlug.org")
http://vip.hex.net/~cbbrowne/resume.html
If the automobile had improved as  much as the computer in the last 25
years, it would get 1,000,000 miles to the gallon, cost $1, have a top
speed of  30,000 mph, and  explode at least  once a year,  killing all
passengers.

------------------------------

From: [EMAIL PROTECTED]
Subject: Re: what 'language' to learn
Date: Wed, 04 Apr 2001 20:15:52 GMT

andi <[EMAIL PROTECTED]> writes:
> I'm new to linux and want to make the best use of it (for admin
> rather than development)

> I'm on a limited budget and don't have huge ammounts of time.

> I'd be interested in learning one of the scripting/coding languages
> (eg C or Perl or somesuch) - but I'd be interested in people's
> opinions of what to learn.

C is useful to know at least a bit of; it's not the prettyest choice
for admin work, as you need to struggle some with pointers and
compilers.

I'd think Perl to be a better choice.  I'd also suggest you look into
Python, which is often considered more elegant than Perl.
-- 
(reverse (concatenate 'string "gro.mca@" "enworbbc"))
http://vip.hyperusa.com/~cbbrowne/resume.html
"Everything should be built top-down, except the first time."
-- Alan Perlis

------------------------------

From: [EMAIL PROTECTED]
Subject: Re: What are servers/services ?
Date: Wed, 04 Apr 2001 20:15:53 GMT

Arctic Storm <[EMAIL PROTECTED]> writes:
> I often read about servers and services, and I think I know what it is, but 
> not fully.

> I'm not talking about someting like a file server or a printer
> server; they obviously allow clients computers to access files on
> hard drives on the host computer, or allow printing.  A host
> computer providing a service to a client computer referrs to a
> network of computers.

> However, if you go to the Hurd section of GNU website, it says that
> Hurd is a collection of servers that run on the Mach microkernel.
> Servers?!  When you talk about the "X server", what is being served?
> If you have a single, isolated computer running Linux & XFree86, how
> can you have a server?  The concept of a host system providing a
> service to a client system cannot apply here; no network.

> How does a server/service pertain to a Daemon?

> Servers/services are also seen in Win2K, so it must have something
> to do with operating systems, in general.

Let's point to X as the canonical example of this; the principles will 
map to other sorts of systems fairly well.

With X, the terminology that you _expected_ gets turned on its ear.

What you were probably _expecting_ would be for an X terminal sitting
on your desk to be considered a "client."  And the big computer that
it connects to, in the back room, would be a "server."

But that's not what happened.

The X terminal runs an X "server."  And programs that connect to that
are called "X clients."

The understanding comes when you phrase things along the lines of
"What is providing services?"

- That X term on your desk is offering graphical services to whatever
  programs come along that want to do graphical "stuff."  So it's a
  "server."

- A program that doesn't work properly without connecting to a
  graphical device goes off requesting graphical services, thus making a
  _client_.

The underlying point is that any _program_ that provides some sort of
service is, in fact, a _server process_.  The client might be another
program that is running on the very same computer; doesn't matter.

A web server is a program that responds to requests from web clients.
They might be running on the same computer; the server doesn't stop
being best termed "server" just because they're on the same computer.

A web server such as Apache might set up connections to pull
information from a database server; in effect, Apache would be
simultaneously behaving as a server _and a client_, based on those
simultaneous roles that it is playing.

I would guess that the thing you need shake out of your head is the
notion that a "server" is primarily tied to being a single-purpose
piece of hardware.  Here, the term is used not to describe how
expensive the case is, but rather the abstraction of the roles being
played by the system components.

CORBA is the canonical example of a technology that stretches this
understanding; a CORBA "server" is any bit of software that accepts
requests for a given IDL interface, whilst a CORBA "client" is any
piece of software that is submitting such requests.  In a CORBA
system, it is highly likely that there will be a mixture of client and
server roles; a chunk of "server" code may very well include requests
to other servers, with a resultant deep interwtining of roles.

With Hurd, the system is constructed as a set of "server" processes
that may well call one another:
  auth, crash, devio, devport, exec, ext2fs, fifo, fwd, ifsock, init,
  magic, new-fifo, nfs, null, pfinet, pflocal, proc, symlink, term,
  ufs

For instance, if you try to read a file, your program might submit a
request to the ext2fs server.  It might then call the "auth" server to
see if you're authorized to access the file.

In contrast, Linux is a "single server;" one big monolithic program
with pieces that call one another, but which are not sufficiently
independent as to be considered "servers."
-- 
(reverse (concatenate 'string "gro.mca@" "enworbbc"))
http://www.ntlug.org/~cbbrowne/resume.html
Necessity  is the  mother  of  invention. Insanity  is  the mother  of
straitjackets.

------------------------------

From: [EMAIL PROTECTED] (Elf Sternberg)
Subject: Re: The Python is so powerful, easy to use? compared with C++
Date: 4 Apr 2001 20:16:51 GMT

In article <i0ty6.11133$[EMAIL PROTECTED]> 
    [EMAIL PROTECTED] (Hun) writes:

>I have a guide book for the Python language. I bought it one month ago,
>but not yet to try learn it.  I read the introduction chapter of the
>book today.  "Five or eight times faster than using C++ to develop a
>application" "Three months length of the Python programming normally
>takes one year of C++ programming."  If so, it would be the ultimate
>programming language.

        I started programming in Python two weeks ago for a new job.
I've been a Perl programmer since 1992.  

        Python is everything it claims to be, much to my amazement.
Yes, it's interpreted, but these days, getting more performance out of a
machine involves throwing more hardware at it, not switching to another
language.  Python is mind-bogglingly easy to learn in and of itself; I
frequently find that what's really hard about yet another procedural
language is learning the library that surrounds it.  Java is notorious
for this; everyone claims "It's just as easy as C++," and it is, until
you're faced with learning the ins and outs of Java's threading model,
or its windowing model, or its I/O model.  

        Python's thread model makes a lot of sense and seems very
straightforward.

        I've also been working with WebWare, the Python Server Page
Application Server.  It took only an hour to hook it into apache, it's
very fast even on my little Pentium-I 333, and it makes so much sense
that I may abandon recommending JSP to any of my customers.

        Python has a database standard and supports Postgres, MySQL, and
Oracle databases.  I've worked with all three.  

        It just blows my mind; I write web clients in minutes, a minimal
Webserver took half an hour, a full-blown Webserver took only a little
longer.  The tools that come with it make Java look second-rate.

>I normally use C and C++. When I read source code, to understand the
>structure and basic ideas is most time consuming not the language
>stuff. C/C++ is so difficult to use? Many developers around me just
>giving up learning C/C++. They prefer Java, script languages or
>4GL. Maybe they are smart or I'm dump head standing out of the
>change. :(

        To be honest, I use Python for prototyping as much as I do for
deployment.  Python's model is close enough to C++ that, to an XP or
UML programmer, it doesn't make much difference.  You can write in
Python and then move on to C++ if you want, although for many of the
programs you might want to write, Python is straightforward.

>Well, I'm considering learning a new language: perl either python. "8
>times faster" still buzzing in my head. :)

        Python and Perl have very different philosophies.  Perl's is
"There's More Than One Way To Do It," and the language reflects this:
there are three different ways of passing arguments to a function, there
are two different ways of representing an array, there are a dozen
different ways of working with a database.  Python, in contrast, has a
more CS-like credo: There's One Obvious and Right Way To Do It.  Without
being a bondage&discipline language like Ada, Python succeeds in making
it simple to do the very hard things, and possible to do the impossible.

        Of course, that's just my opinion after working with the
language for two weeks.  I may change my mind again.

                Elf

--
Elf M. Sternberg, rational romantic mystical cynical idealist
http://www.halcyon.com/elf/

Fast food restaurants are like gay bathhouses in San Francisco, 
places where people go to engage in high-risk behaviors.
                - Greg Critser

------------------------------

From: Lynne Lawrence <[EMAIL PROTECTED]>
Subject: Re: Startup items in RH6.2
Date: Wed, 04 Apr 2001 16:34:21 -0400

Sizam wrote:

     How do I find what programs get turned on at startup?  My server went

     down during a blackout and came back on but didn't startup sendmail
     and
     started up the wrong apache program, where can I change this?

     Thanks,

Sam

   * check your /etc/inittab file for your system's default runlevel
   * go to directory /etc/rc<n>.d, where <n> is the default run level
   * the files that start with an S will be run with a 'start' option when
     this run level is entered.  They typically are symbolically linked to
     files in /etc/init.d (where there are probably scripts that do what
     you want).
   * Find the script in /etc/init.d that starts the app (i.e. sendmail),
     and link it to an S file in the appropriate rc<n>.d file (ln -s
     ../init.d/sendmail S80sendmail').  The S files are executed in order,
     so make sure your networking and such is started before sendmail.


------------------------------

From: hoffmyster <[EMAIL PROTECTED]>
Subject: Re: I would like to register a complaint ...
Date: Wed, 04 Apr 2001 16:36:00 -0400

I do appreciate the responses to my supposed dilemma of trying to "format" a
partition in Linux.

I now realize that there IS NO equivelent in Linux.  The reason for this
apparently is because the ext2 fs is so much of a different animal than the FAT
fs.  The most obvious differnce is the use of superblocks by ext2.

The DOS format command destroys everything then rebuilds the fs with whatever
it needs.

In Linux it appears the closest thing is rm -rf *.   This wipes out all files
and directories under the current directory without touching the filesystem,
ie. superblocks, inodes, etc.  Then you can write whatever you want back into
the "empty" partition right away.  Of course you wouldn't want to do this to a
partition on the installation you have booted up in and are currently running.
I am talking about "formatting" a partition on totally another hard drive that
I mounted onto the current system.

By the way:

The "dd if=/dev/zero of=/dev/hdxX" or "cat /dev/zero /dev/hdxX" IS NOT the way
to do this because not only does it wipe out the superblocks but fills the
entire partition with one huge file full of zeros that involves entirely
something else to get rid of.  BAD ADVISE, ERIC!!

mke2fs will restore the superblocks and build the inodes but will not destroy
any data, so it by itself is not "formatting" a partition.

Actually in Linux as it turns out doesn't need a "format" command because the
mounted partition is in itself is a directory and you can just clear its data
by deleting it files and subdirectories.  The fs remains intack.

I'll take back my complaint now but it sure was a learning experience trying to
destroy my own data :-)

RLH


------------------------------

From: "Nelson Yip" <[EMAIL PROTECTED]>
Subject: Dhcp
Date: Wed, 4 Apr 2001 15:47:48 -0400

Hi everyone,

I have a big problem here...I configured dhcp v3.0b2118
when I type dhcpd start.....I get a message stating no defined subnet
(0.0.0.0)
here is my dhcpd.conf file
Please tell me am I going crazy or there is a bug in this beta version

# Define global values that apply to all systems.
#
option domain-name "igicom.com";
option domain-name-servers 192.168.168.43;
#option netbios-name-servers ;
#option netbios-nod-type 8;
#option nis-domain "igicom";
#
# Set up a subnet
#
subnet 192.168.168.0 netmask 255.255.255.0 {
        range 192.168.168.100 192.168.168.254;
        default-lease-time 604800;
        max-lease-time 604800;
        option subnet-mask 255.255.255.0;
        option broadcast-address 192.168.168.255;
        option routers 192.168.168.1;
        }
#
ddns-hostname "jukebox2.igicom.com";
ddns-update-style ad-hoc;

--
Nelson Yip
System Administrator
Igicom, Inc.





------------------------------

From: [EMAIL PROTECTED] (Juergen Heinzl)
Subject: Re: what 'language' to learn
Date: Wed, 04 Apr 2001 20:55:17 GMT

In article <[EMAIL PROTECTED]>, andi wrote:
>I'm new to linux and want to make the best use of it (for admin rather
>than development)
>
>I'm on a limited budget and don't have huge ammounts of time.
>
>I'd be interested in learning one of the scripting/coding languages
>(eg C or Perl or somesuch) - but I'd be interested in people's
>opinions of what to learn.
[-]
Shell, Perl and the rest like sed, awk and those numerous
utilities.

If you're going to work with other Unices too, then sed and
awk can become more important. Yes, Perl is pretty standard
nowadays, still you may find rather old Unix installations
happily humming along.

See http://www.ora.com/ for literature.

Ta',
Juergen

-- 
\ Real name     : Juergen Heinzl                \       no flames      /
 \ EMail Private : [EMAIL PROTECTED] \ send money instead /

------------------------------

From: hoffmyster <[EMAIL PROTECTED]>
Subject: Using RPM with RH5.2
Date: Wed, 04 Apr 2001 16:56:10 -0400

Well, I am now onto my next learning experience/fight :-) with Linux.

I am simply trying to upgrade packages in my RH5.2.  Whenever I try
using "rpm -U <rpm name>" I get this message:

"only packages with major numbers <=3 are supported by this version of
RPM
  package <rpm name> can not be installed"

Does anyone know what this means??  How do you upgrade a version of
RPM.  I tried upgrading to the latest 4.0.2 but I continue to get this
message.

My current RPM version is 3.0.1.  I am using egcs 2.91.66 with GNU make
3.77.  There is a packages.rpm in my /var/lib/rpm directory.
I can perform queries and stuff but I can't get it to upgrade/install
any new RPM's.

Is anybody familiar with this problem?  What am I doing wrong?

Thanks.

RLH


------------------------------

From: "Eric Potter" <[EMAIL PROTECTED]>
Subject: Re: CDR writing with SCSI emulation stops ppp transfers
Crossposted-To: alt.os.linux.mandrake
Date: Wed, 04 Apr 2001 20:57:48 GMT

In article <[EMAIL PROTECTED]>, "Esa Tikka"
<[EMAIL PROTECTED]> wrote:

> On Tue, 03 Apr 2001 17:01:47 GMT, Eric Potter <[EMAIL PROTECTED]>
> wrote:
>>
>>This is a weird one, but whenever I write a disk with my IDE CDRW, any
>>downloads via my dial-up modem stop immediately.  If I stop writing the
>>CDR, then the download resumes.  I have tried this on two different
>>computers, and two different distributions, Mandrake 7.2 and Mandrake
>>8.0b2.  I have tried my own kernel builds also.  CD writer is set as
>>secondary master.  This only seems to occur when using the ide-scsi
>>emulation for atapi CDR's.  If I don't load ide-scsi, I can run "cat
>>/dev/hdc > /dev/null" or another such heavy I/O operation without this
>>problem.
> 
> It would seem that this is not the case, but still... have you tried
> setting unmasqirq=on on your ide drives?  (hdparm -u1 /dev/hdx) Read the
> man page of hdparm first anyway, there are some warnings worth noting.
> I had similar problems, stressing *any* ide drive caused serial port to
> stop transmitting data.
> 
> Just grasping straws...
> 
> 

unmaskirq = 1  is already set for all IDE devices.

So I set 'hdparm -u0 /dev/hdc' and tried again, but the same result.

Also went into bios setup and reserved IRQ 3 and 4 for legacy ISA.
Modem is external on com2 irq 3.



------------------------------

From: "Eric Potter" <[EMAIL PROTECTED]>
Subject: Re: CDR writing with SCSI emulation stops ppp transfers
Crossposted-To: alt.os.linux.mandrake
Date: Wed, 04 Apr 2001 20:57:56 GMT

In article <[EMAIL PROTECTED]>, "David Efflandt"
<[EMAIL PROTECTED]> wrote:

> On Tue, 03 Apr 2001 17:01:47 GMT, Eric Potter <[EMAIL PROTECTED]>
> wrote:
>>In article <3ac92dd8$0$42878$[EMAIL PROTECTED]>, "Dances With
>>Crows" <[EMAIL PROTECTED]> wrote:
>>
>>> On Mon, 02 Apr 2001 15:58:38 GMT, Eric Potter staggered into the Black
>>> Sun and said:
>>>>This is a weird one, but whenever I write a disk with my IDE CDRW, any
>>>>downloads via my dial-up modem stop immediately.  If I stop writing
>>>>the CDR, then the download resumes.  I have tried this on two
>>>>different computers, and two different distributions, Mandrake 7.2 and
>>>>Mandrake 8.0b2.  I have tried my own kernel builds also.  CD writer is
>>>>set as secondary master.  This only seems to occur when using the
>>>>ide-scsi emulation for atapi CDR's.  If I don't load ide-scsi, I can
>>>>run "cat /dev/hdc > /dev/null" or another such heavy I/O operation
>>>>without this problem.
>>> 
>>> How are you writing this CD?  cdrecord, or one of its front-ends?  And
>>> do dmesg or /var/log/messages show anything odd happening when you
>>> start to burn the CD?  cdrecord may try to set itself to near-realtime
>>> priority (nice -19 or so) when it shouldn't.  (Check the output of
>>> "top" while cdrecord is running.)  Frontends like XCDRoast may have a
>>> switch you can flip to make cdrecord run at normal priority.  If
>>> cdrecord is hogging the CPU, then said CPU will be hard-pressed to
>>> handle the near-constant stream of IRQs that a modem generates.
>>> 
>>> Also check what "cat /proc/interrupts" says regarding the secondary
>>> IDE scontroller.  Some drives generate a lot more IRQs than others. 
>>> HTH,
>>>
>>>
>>Thanks for some good suggestions.  I am using cdrecord.
>>
>>cdrecord -v -dummy speed=8 fs=8m dev=2,0,0 /dev/scd1
>>
>>It doesn't matter what speed I am using, 2, 4, or 8 I get the same
>>problem.  /var/log/messages shows nothing at all about the problem. It
>>also doesn't matter where the input file is stored,  IDE or SCSI hard
>>disk or SCSI CDROM.  'Nice' value is 0 for the process, and I am using a
>>1 Ghz athlon, so cpu usage is not a problem.  Here are IRQs:
>>           CPU0
>>  0:     539582          XT-PIC  timer
>>  1:       2103          XT-PIC  keyboard 2:          0          XT-PIC 
>>  cascade 3:    3131334          XT-PIC
>>  7:          0          XT-PIC  parport0
>> 10:       2833          XT-PIC  ncr53c8xx 11:         49         
>> XT-PIC  BusLogic BT-948, es1371 12:     220819          XT-PIC  PS/2
>> Mouse 14:      27372          XT-PIC  ide0
>> 15:       2752          XT-PIC  ide1
>>NMI:          0
>>ERR:          0
> 
> Strange that your irq 3 does not show what is using it, should say
> serial. It is recognized by the kernel below.  What kind of modem is it
> (ISA, PNP, PCI, external)?  My external modem on ttyS0 shows up like
> this:
> 
>   4:     135803          XT-PIC  serial
> 
> I read somewhere that fast scsi can sometimes steal interupts because it
> tries to do its thing fast without using them.  Although, I would think
> that would just affect something using the same irq.

My modem is external on com2 IRQ 3.  I just reserved IRQs 3 and 4 for 
legacy ISA in my bios setup, but that didn't help.

> 
>>Here is dmesg:
>>
>>Serial driver version 5.05 (2000-12-13) with MANY_PORTS SHARE_IRQ
>>SERIAL_PCI enabled ttyS00 at 0x03f8 (irq = 4) is a 16550A ttyS01 at
>>0x02f8 (irq = 3) is a 16550A
> (snipped)
>







------------------------------

From: "Eric Potter" <[EMAIL PROTECTED]>
Subject: Re: CDR writing with SCSI emulation stops ppp transfers
Crossposted-To: alt.os.linux.mandrake
Date: Wed, 04 Apr 2001 20:57:54 GMT

In article <3aca71c4$0$42877$[EMAIL PROTECTED]>, "Dances With
Crows" <[EMAIL PROTECTED]> wrote:

> On 3 Apr 2001 17:17:53 GMT, Esa Tikka staggered into the Black Sun and
> said:
>>On Tue, 03 Apr 2001 17:01:47 GMT, Eric Potter <[EMAIL PROTECTED]>
>>wrote:
>>>
>>>This is a weird one, but whenever I write a disk with my IDE CDRW, any
>>>downloads via my dial-up modem stop immediately.  If I stop writing the
>>>CDR, then the download resumes.  I have tried this on two different
>>>computers, and two different distributions, Mandrake 7.2 and Mandrake
>>>8.0b2.  I have tried my own kernel builds also.  CD writer is set as
>>>secondary master.  This only seems to occur when using the ide-scsi
>>>emulation for atapi CDR's.  If I don't load ide-scsi, I can run "cat
>>>/dev/hdc > /dev/null" or another such heavy I/O operation without this
>>>problem.
>>
>>It would seem that this is not the case, but still... have you tried
>>setting unmasqirq=on on your ide drives?  (hdparm -u1 /dev/hdx) Read the
>>man page of hdparm first anyway, there are some warnings worth noting.
>>I had similar problems, stressing *any* ide drive caused serial port to
>>stop transmitting data.
> 
> What Esa Tikka said.  hdparm -m16 -u1 -c1 /dev/hdX is safe on just about
> any modern hardware; adding -d1 is a bit riskier but may improve
> throughput by a lot.
> 
> Also, you should've mentioned that you were using kernel 2.4.3 . Despite
> its even number in the second position, there are still some bugs. 
> IIRC, cdrecord gets some of its vital info from kernel headers, and if
> you're using a cdrecord compiled against a 2.2 kernel while using a 2.4
> kernel, odd problems may crop up.
> 

Just recompiled cdrecord against 2.4.3 kernel, and I get the same result.
I reserved IRQs 3 and 4 in bios setup, so I'm thinking that it may be a 
kernel issue, not an IRQ issue.  I will go back to 2.2 kernel and try
again, but first I have to switch some hardware, as I am using Reiserfs
right now.  Will post results tomorrow.








------------------------------

From: Markku Kolkka <[EMAIL PROTECTED]>
Subject: Re: Startup items in RH6.2
Date: 04 Apr 2001 23:30:36 +0300

Sizam <[EMAIL PROTECTED]> writes:

> How do I find what programs get turned on at startup?

chkconfig --list

>  My server went 
> down during a blackout and came back on but didn't startup sendmail and 
> started up the wrong apache program, where can I change this?

chkconfig <service> on
chkconfig <service> off

See man 8 chkconfig for details. There are many more commands to
manage startup configuration, e.g. ntsysv, tksysv, linuxconf, webmin,
etc. If your system is confused you may have to edit files in
/etc/rc.d manually.

-- 
        Markku Kolkka
        [EMAIL PROTECTED]

------------------------------

From: Jeremiah DeWitt Weiner <[EMAIL PROTECTED]>
Subject: Re: I would like to register a complaint ...
Date: 4 Apr 2001 21:01:01 GMT

hoffmyster <[EMAIL PROTECTED]> wrote:

> The "dd if=/dev/zero of=/dev/hdxX" or "cat /dev/zero /dev/hdxX" IS NOT the way
> to do this because not only does it wipe out the superblocks but fills the
> entire partition with one huge file full of zeros that involves entirely
> something else to get rid of.  

        No, it doesn't.  It fills the partition full of zeros.  It does
not fill it with "one huge file full of zeros".  Since you have destroyed
the metadata that makes it a filesystem, nothing in there is a "file" in 
any meaningful sense any more.

        Observe:
chindi:~$ sudo mke2fs /dev/fd0
mke2fs 1.15, 18-Jul-1999 for EXT2 FS 0.5b, 95/08/09
(regular mke2fs output)

chindi:~$ sudo dumpe2fs /dev/fd0
dumpe2fs 1.15, 18-Jul-1999 for EXT2 FS 0.5b, 95/08/09
Filesystem volume name:   <none>
Last mounted on:          <not available>
Filesystem UUID:          7216ecce-e8cd-44dc-b576-5d01c558bda8
(etc.  It's a filesystem, with regular filesystem metadata.)

chindi:~$ sudo dd if=/dev/zero of=/dev/fd0
dd: /dev/fd0: Input/output error
1009+0 records in
1008+0 records out
(Sayonara, filesystem.)

chindi:~$ sudo dumpe2fs /dev/fd0
dumpe2fs 1.15, 18-Jul-1999 for EXT2 FS 0.5b, 95/08/09
dumpe2fs: Bad magic number in super-block while trying to open /dev/fd0
Couldn't find valid filesystem superblock.

        Look at that...no more files, no nothing!  Certainly not "one
huge file full of zeros".  While this may not be the behavior you wanted,
it is undeniably one way to get rid of the files in the filesystem, and
it does not require anything else to get rid of it either, although you
will have to remake the filesystem.  The difference is that the dd
method actually overwrites the medium, whereas just a new mke2fs will
make the files disappear from the filesystem, but not the medium.  You
owe Eric an apology; there's nothing wrong with his advice.

JDW



------------------------------

From: Dranthony <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Subject: Re: missing initab file.......enter runlevel?
Date: Wed, 04 Apr 2001 21:02:46 GMT

My fs is messed.  everything in /etc directory is messed so no matter
how I spel it, it won't work.


------------------------------


** 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 by posting to comp.os.linux.misc.

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
******************************

Reply via email to