Linux-Misc Digest #714, Volume #26 Fri, 5 Jan 01 13:13:01 EST
Contents:
Re: auto run ([EMAIL PROTECTED])
Re: Installing on laptop... (Dave Brown)
Re: wu-ftp question ("Peter T. Breuer")
Re: Compiling the Linux Kernel ("Peter T. Breuer")
Re: auto run ("Peter T. Breuer")
Re: Switching to Linux (Lori Holder-Webb)
Re: Why did SawFish become default in RedHat, instead of Enlightenment ? (Lori
Holder-Webb)
cp -i -f ([EMAIL PROTECTED])
Re: Red Hat dead/dying? (John A. Stewart)
ipchains question ("Londonboy")
Verify Email address? ("frank")
Re: accton (Eggert Ehmke)
upgrade 2.4.0, init: Id "c1" respawning too fast ("Duke")
Linux 2.4 on RH7.0 ([EMAIL PROTECTED])
Re: Red Hat dead/dying? ("Peter T. Breuer")
Very Strange (and Severe) Linux Crash ([EMAIL PROTECTED])
Re: Any way to specify which eth a service listens to? (Xirphoid)
Re: Linux 2.4 on RH7.0 ("Chip Piller")
Re: ipchains question (David)
Re: Linux Browsers ("Bill Piety")
Re: Switching to Linux ("Peter T. Breuer")
Adding a module when I've only got the binary ([EMAIL PROTECTED])
Re: who's rewriting /etc/fstab? ("Chip Piller")
----------------------------------------------------------------------------
From: [EMAIL PROTECTED]
Subject: Re: auto run
Crossposted-To: comp.os.linux.advocacy
Date: Fri, 5 Jan 2001 13:53:27 +0000
Aaron R. Kulkis <[EMAIL PROTECTED]> did eloquently scribble:
> ZippiZ wrote:
>>
>> I am using Turbolinux and Debian, I know the way to set the program to
>> autorun at startup is the rc.local in Turbolinux, but what is that in
>> Debian? Thanks for any suggestion!
>>
>> ZippiZ
> What the fuck is autorun?
Lets look at the word more closely shall we...
First, lets split it in two...
Auto... Short for automatic. To do something without manual intrervention.
Run... To start a program...
So to autorun a program means to start a program auotomatically...
</sarcasm>
Can we say Duuuuh???
Zippy just didn't mention what program he wants to autorun.
I've never used Turbo or Debian, so I don't know if it uses sysv init style
startups.
Oh... And for a unix system enginner, don't you think you should KNOW about
such things a sig file size limits???
Mine's a couple of lines too long, but you're taking the piss with that one!
--
______________________________________________________________________________
| [EMAIL PROTECTED] | |
|Andrew Halliwell BSc(hons)| "The day Microsoft makes something that doesn't |
| in | suck is probably the day they start making |
| Computer science | vacuum cleaners" - Ernst Jan Plugge |
==============================================================================
------------------------------
From: [EMAIL PROTECTED] (Dave Brown)
Subject: Re: Installing on laptop...
Reply-To: [EMAIL PROTECTED]
Date: 5 Jan 2001 10:21:40 -0600
In article <9341n4$2gr$[EMAIL PROTECTED]>, [EMAIL PROTECTED] wrote:
>I have a Pentium 133 mhz with 16 megs of ram
>laptop. What would the best linux to install on
>it.
>I dont know linux very well but I do have some
>experience with Mandrake 6.2 but that is it. I
>also have it on cd already and would it run ok
>on it using KDE ?
Needless to say, it depends on what you want to do with it. But,
"we" used run on a 486 with 8 megs of memory, and could even
run X, although 16 megs sure came in handy. With "Linux" trying
to "catch up" with Win, more and more apps are memory hogs. Load
up enlightenment and Netscape and you're probably already swapping
on a 32 meg machine.
I still use Slackware 3.3 (partially upgraded to 3.9) on a laptop
for travel. 2.0.38 kernel is fast and efficient (for what it could
do). One of the fvwm-derivative windows managers does the
job with minimal memory usage. Runs Netscape, WordPerfect7 for
Linux, Acrobat Reader, Freecell, (not necessarily at the same time),
etc.
--
Dave Brown Austin, TX
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: wu-ftp question
Date: Fri, 5 Jan 2001 17:15:54 +0100
Tom Edelbrok <[EMAIL PROTECTED]> wrote:
> My wu-ftp works when I open someone elses ftp site but no one can ftp in to
wuftpd is a server. I don't know what client you have.
Peter
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Compiling the Linux Kernel
Date: Fri, 5 Jan 2001 17:14:20 +0100
[EMAIL PROTECTED] wrote:
> : tools/build.c:25: stdio.h: No such file or directory
You're missing all the C headers. Instead of asking here, use rpmfind
to locate what you need (or the MANIFEST of your distro).
Peter
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: auto run
Crossposted-To: comp.os.linux.advocacy
Date: Fri, 05 Jan 2001 16:35:03 GMT
In comp.os.linux.misc [EMAIL PROTECTED] wrote:
> Aaron R. Kulkis <[EMAIL PROTECTED]> did eloquently scribble:
> First, lets split it in two...
> Auto... Short for automatic. To do something without manual intrervention.
"Auto" means "self" as far as I know! As in "automobile" = self-moving.
> Run... To start a program...
Look up "locofoco" in the dictionary. A good one.
Peter
------------------------------
From: Lori Holder-Webb <[EMAIL PROTECTED]>
Subject: Re: Switching to Linux
Date: Fri, 05 Jan 2001 10:22:39 -0600
I _did_ say "ugh" about csh. That is the default shell for my mainframe
account, and what I don't like about it is the lack of command-line
editing.
Bash does fine by me, I'm not a power-scripter. Mostly I like the
command line editing, the <TAB> command completion, the access to
command line history, etc. Stuff I couldn't get from the csh
implementation on the mainframe...
"Peter T. Breuer" wrote:
>
> Lori Holder-Webb <[EMAIL PROTECTED]> wrote:
> > after I switched my mainframe shell from csh (UGH!) to bash, the
>
> Why would anyone say "ugh" about tcsh? Or are you really meaning "csh"?
> In that case, I can understand it, because csh is to tcsh as sh is to
> bash. You should use tcsh (it's STILL a much better shell than bash,
> despite bashes years of playing at catch-up).
>
> Peter
------------------------------
From: Lori Holder-Webb <[EMAIL PROTECTED]>
Subject: Re: Why did SawFish become default in RedHat, instead of Enlightenment ?
Date: Fri, 05 Jan 2001 10:28:42 -0600
How is Sawfish broken? I've been using it for a year now and have had
no problems at all. It hasn't segfaulted once (compared to E, which
segfaulted at least twice a month on the same system). It hasn't locked
up. It starts up and runs faster than E.
My comparison was based on running both w-m on a K6-2 450 MHz system
with 64 MB RAM. Perhaps E is not such a problem on a faster box, but
likewise, Sawfish ought to *fly* in that case...
I decided to jump the E-ship for good when the new version (of E) came
out and I discovered that it had become larger, less easy to configure,
and replicated more of the gnome features with which I was already
satisfied.
Silviu Minut wrote:
>
> > RedHat is a strong supporter of Gnome. The initial release of Gnome
> > recommended Enlightenment as it's window manager. However there were
> > many problems. Especially with the fact that both Gnome and
> > Enlightenment wanted to provide the same set of features. A
> > light-weight window manager meant to be used with gnome was needed.
> > That window manager was Sawfish (orginally Sawmill). Sawfish is now
> > the "gnome" windows manager. It's a much better fit for gnome than
> > enlightenment ever was.
>
> Except that Sawfish is a little broken...
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: alt.os.linux
Subject: cp -i -f
Date: Fri, 05 Jan 2001 16:37:28 GMT
hi
Why is it that
/bin/cp -i -f
prompts when overwriting, while
/bin/mv -i -f
and
/bin/rm -i -f
do not prompt?
This seems inconsistent.
cp (GNU fileutils) 4.0x
Wroot
Sent via Deja.com
http://www.deja.com/
------------------------------
From: [EMAIL PROTECTED] (John A. Stewart)
Crossposted-To: comp.os.linux.advocacy,comp.unix.solaris
Subject: Re: Red Hat dead/dying?
Date: 5 Jan 2001 16:47:02 GMT
In <[EMAIL PROTECTED]> "Aaron R. Kulkis" <[EMAIL PROTECTED]> writes:
>Replace Windows with Linux....
>Seriously. I recently did a contract with EDS. We supported over
>5,000 Unix workstations at GM sites scattered all over Michigan and
>even farther away from one room with a staff of less then 20 people.
>This took care of all problems other than hardware failures.
>For those situations, each GM site has a couple of people on duty
>during business hours to do things like keyboard/mouse/spaceball
>replacement, and escorting the vendors' Service Engineers to the
>workstation in question if more extensive work needed to be done.
>These people are barely Unix-literate.
I'm a Unix system administrator but I can see that most of our users are
better served with a Windows machine on their desktop. It's the
applications and common look and feel that make this the only realistic
option for the desktop. I've seen a local telecom manufacturer that still
provided Sun workstations to their engineers, but a large portion of their
computer room was occupied with NT servers that enabled the same
engineers to run Word, Excel, Powerpoint and other popular windows
applications. They have since switched to Microsoft Exchange mail.
When I see an organization become more and more dependent on Windows
applications even though they claim to be primarily a Unix shop I know it
is only a matter of time before they wake up and realize that it would
be better to just drop a Windows box on the desktop so all these
applications can be run in native mode.
Our experience is that Windows machines are easier to support in large
numbers. We can simultaneously drop a new image on hundreds of machines
using Norton Ghost mostly with the effort of a few junior staff members.
The much smaller number of Sun workstations we support are much more
difficult to manage because of the tendancy for Unix systems to be
configured to provide services and for machines to be kept in service much
longer than Windows machines. We've tried to centralize services as much
as we can and force people to retire older machines but the effort and
level of expertise required to support these machines is still high relative
to what it takes to support Windows desktop machines.
I should also mention that most universities also suffer from an
empire building mentality. Instead of providing basic services such as
email and file service on a few large well maintained servers, you find
many departments attempting to support several different operating systems
and associated servers with only one or two staff members.
--
John Stewart -- Computing and Communications Services, Carleton University
Internet: [EMAIL PROTECTED] 613-520-2600x3707
"you are incompetent bloodlusting idiot pirates. -- CCS student feedback"
------------------------------
From: "Londonboy" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.redhat,linux.redhat.misc
Subject: ipchains question
Date: Fri, 05 Jan 2001 16:56:23 GMT
I am trying to tighten the security of my server by using IPCHAINS, and I
read a lot of articles on the web about that. There are 2 methods:
1. Some of them don't suggest change the INPUT policy to DENY, it should
remind ACCEPT, and then block/open all the "known" ports.
2. There are articles suggest INPUT policy should be set to DENY first, and
then open necessary ports.
So I followed the instructions of (2) but I experienced a lot of problems,
for example FTP didn't work, SSH login become very slow, strange!! Then I
switched to the frist method, everything works fine.
Can anyone tell me why the 2nd one doesn't work well??
Thanks
Linux Newbie
------------------------------
From: "frank" <[EMAIL PROTECTED]>
Subject: Verify Email address?
Date: Fri, 05 Jan 2001 16:56:29 GMT
Hi:
Is there an easy way or a simple script that can verify email address in a
mailing list?
Thank you,
Frank
------------------------------
From: Eggert Ehmke <[EMAIL PROTECTED]>
Subject: Re: accton
Date: Fri, 05 Jan 2001 18:01:43 +0100
Reply-To: [EMAIL PROTECTED]
On Fri, 05 Jan 2001 09:30:09 GMT, Speedlnx <[EMAIL PROTECTED]> wrote:
>In my linux redhat 6.2 kernel 2.2.17 at every shutdown a process fail with the
>message: "accton: Funzione non implementata"
>Can someone tell me why it fail and how to fix this problem?
Accton is a manufacturer of Network Interface Cards (NIC's). Do you have
such a card ? Is there some option in your kernel that is related to that ?
My Accton card works with the NE2000 driver.
Eggert
--
Eggert Ehmke
Email: [EMAIL PROTECTED]
------------------------------
From: "Duke" <[EMAIL PROTECTED]>
Subject: upgrade 2.4.0, init: Id "c1" respawning too fast
Date: Fri, 5 Jan 2001 12:01:36 -0500
Just upgraded my 2.2.18/glibc-2.1.3 machine to kernel 2.4.0 and am now
getting this in the syslog every 5 minutes:
Jan 5 11:49:02 hobo init: Id "c1" respawning too fast: disabled for 5
minutes
Jan 5 11:49:02 hobo init: Id "c3" respawning too fast: disabled for 5
minutes
Jan 5 11:49:02 hobo init: Id "c2" respawning too fast: disabled for 5
minutes
Jan 5 11:49:02 hobo init: Id "c4" respawning too fast: disabled for 5
minutes
Jan 5 11:49:03 hobo init: Id "c5" respawning too fast: disabled for 5
minutes
Jan 5 11:49:03 hobo init: Id "c6" respawning too fast: disabled for 5
minutes
Have not changed anything else. Any ideas why this would be?
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: alt.os.linux
Subject: Linux 2.4 on RH7.0
Date: Fri, 05 Jan 2001 17:03:31 GMT
Hi,
People who are already using the new kernel on RH7.0, how is it? Is
RH7.0 really 2.4 ready? Is it worth compiling or should I wait for the
next official RH or Mandrake version?
Thanks
Wroot
Sent via Deja.com
http://www.deja.com/
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Red Hat dead/dying?
Crossposted-To: comp.os.linux.advocacy,comp.unix.solaris
Date: Fri, 05 Jan 2001 17:25:04 GMT
In comp.os.linux.misc John A. Stewart <[EMAIL PROTECTED]> wrote:
> Our experience is that Windows machines are easier to support in large
You must be joking!
> numbers. We can simultaneously drop a new image on hundreds of machines
> using Norton Ghost mostly with the effort of a few junior staff members.
ghost will do the same for any partition, including linux
ones. It's simply a broadcast protocol. and doesn't depend on windows
in any way!
> The much smaller number of Sun workstations we support are much more
There's no reason why they shouldn't all be configured identically.
tftp and NIS+ takes care of the rest.
> difficult to manage because of the tendancy for Unix systems to be
> configured to provide services and for machines to be kept in service much
> longer than Windows machines. We've tried to centralize services as much
> I should also mention that most universities also suffer from an
> empire building mentality. Instead of providing basic services such as
> email and file service on a few large well maintained servers, you find
> many departments attempting to support several different operating systems
> and associated servers with only one or two staff members.
True, but that's not empire building. Empire building is when ONE
department starts to offer a special service for the whole uni,
usurping other departments efforts. An example would be taking control
of broadcast tunnels for video and audio communications.
Peter
------------------------------
From: [EMAIL PROTECTED]
Subject: Very Strange (and Severe) Linux Crash
Date: Fri, 05 Jan 2001 17:20:29 GMT
Reply-To: [EMAIL PROTECTED]
I just thought I'd share my experiences with a very damaging and strange
disk
crash I just had.
First: some architecture:
/dev/hda <-- Big 30Gb Disk with all my valuable stuff
/dev/hdc <-- "Little" 13Gb Disk with MP3s on it.
History:
Over the past few weeks, I had noticed a few problems in the syslog with
errors on /dev/hdc. Since it was only full of MP3s, I wasn't that
worried
about it. Figured I'd get to it later. It did cause me to have to
reboot
a couple of times though, when it stopped responding.
But, yesterday afternoon, things got really bad.
I first noticed I had problems when I tried to send an e-mail. Perl (my
interim sendmail filter thingie is in perl) was failing telling me that
there
was a problem with ld.so. What?!?!? So, I look around, and about half
my
executiables on my good disk are having problems.
I start diging through the logs, assuming I have a hacker. That's when
I saw
the sense problems on hdc, that occured just before my problems. "No
biggie,"
I thought, and just rebooted the box.
But, it didn't come back up. Couldn't find init. After several
attempts, I
assumed I read the error wrong and the big disk (hda) had failed. So, I
start
sweatting, "Just how good are my backups?!?!?"
I ran to frys, bought another disk, and started salvaging. At first
glance,
hda1 looked fine. All partitions were visible. There were no errors
booting,
the disk was quiet . . Awesome! Just re-install the OS and I'm fine.
So, after installing linux onto my new 40gb drive, I started copying
files
over. I knew I was going to have troubles when I saw all the fsck
errors that
needed my confirmation. I started typeing "no" to be safe, and decided
to
access the disk without fscking the partitions. (I checked one, and
there were
literally hundreds of wierd errors)
So, now, I've recovered probably 90% of my data. (Of course, the 10%
was pretty
important stuff :(
I can only assume that there is a bug in the disk driver in linux. That
when
it has problems on one disk, it gets into an error state and starts
corrupting
all disks. The nature of the corruption seems to be the
inodes/directory
information.
Several files had their types changed. (It's not good when a directory
goes
from type "d" to type "b". It's hard to list the contents of a block
special files). I had directories and files become type "b" with a
wierd
uid/gid. I had some files become symbolic links. (Although when they
were ascii files, they pointed to the original file contents, so you
could
ls -l BAD FILE >repairme.txt and manually fix it)
So, to make a long story short: Never ignore disk errors, even on disks
you
don't care about.
Still restoring stuff,
Dave
Sent via Deja.com
http://www.deja.com/
------------------------------
From: Xirphoid <[EMAIL PROTECTED]>
Subject: Re: Any way to specify which eth a service listens to?
Date: Fri, 05 Jan 2001 17:30:09 -0000
In any version of eggdrop, you can define a 'VHOST' in the CONF for
example:
set my-hostname "localhost"
set my-ip "127.0.0.1"
---
At least fill my-ip with the IP of the ethX interface you want it to listen
on, and then it will listen on that interface only.
Fester wrote:
>
>
> I have two cards in my Redhat 7 box. eth0 goes out to the Internet, and
> eth1 goes in to my home network. Services such as samba, and my eggdrop
> bot should only listen on the internal network, and not to everyone. Is
> there any way to tell a service to listen to only a specific ethernet
> adapter?
>
> --
> -- Fester
>
> "Sometimes I may appear to [suck], but it's just a superficial
> blemish on my rulingness." - Sunshine
> =================================================================
>
>
--
Posted via CNET Help.com
http://www.help.com/
------------------------------
From: "Chip Piller" <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux
Subject: Re: Linux 2.4 on RH7.0
Date: Fri, 5 Jan 2001 12:35:00 -0500
I am running RedHat 7.0 with all updated packages on a PIII machine. I
downloaded the 2.4.0-prerelease file, configured, compiled, and installed.
It has been running for several days with no problems whatsoever. I have
not had the opportunity to try out any of the new features though.
Good Luck,
Chip
<[EMAIL PROTECTED]> wrote in message news:934uos$pq6$[EMAIL PROTECTED]...
> Hi,
>
> People who are already using the new kernel on RH7.0, how is it? Is
> RH7.0 really 2.4 ready? Is it worth compiling or should I wait for the
> next official RH or Mandrake version?
>
> Thanks
>
> Wroot
>
>
> Sent via Deja.com
> http://www.deja.com/
------------------------------
From: David <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.redhat,linux.redhat.misc
Subject: Re: ipchains question
Date: Fri, 05 Jan 2001 17:37:49 GMT
Londonboy wrote:
>
> I am trying to tighten the security of my server by using IPCHAINS, and I
> read a lot of articles on the web about that. There are 2 methods:
>
> 1. Some of them don't suggest change the INPUT policy to DENY, it should
> remind ACCEPT, and then block/open all the "known" ports.
>
> 2. There are articles suggest INPUT policy should be set to DENY first, and
> then open necessary ports.
>
> So I followed the instructions of (2) but I experienced a lot of problems,
> for example FTP didn't work, SSH login become very slow, strange!! Then I
> switched to the frist method, everything works fine.
>
> Can anyone tell me why the 2nd one doesn't work well??
You should set the "Default policy to DENY" and only open what you need
to for your system. At the link below is a base firewall I started from
and then added several rules to it.
http://www.redhat.com/support/docs/tips/firewall/firewallservice.html
Or you can go to this one and create one. Must enable java script for it
to work.
http://linux-firewall-tools.com/linux/firewall/index.html
--
Confucius say: He who play in root, eventually kill tree.
Registered with the Linux Counter. http://counter.li.org
ID # 123538
Completed more W/U's than 98.982% of seti users. +/- 0.01%
------------------------------
From: "Bill Piety" <[EMAIL PROTECTED]>
Subject: Re: Linux Browsers
Date: Fri, 05 Jan 2001 11:41:15 -0600
In article <72r46.34793$[EMAIL PROTECTED]>, "Netcop"
<[EMAIL PROTECTED]> wrote:
> Try the mozilla web browser. Netscape 6.0 was built off mozilla
> milestone
> 18, but mozilla realised a updated version mozilla 0.6 what is a lot
> better.
>
> [EMAIL PROTECTED] wrote:
>
>> Hi
>>
>> Why is it that it takes konqueror about twice as long to draw a jpeg
>> picture (stored locally) on the screen as netscape?
>>
>> Why does netscape take up almost all available CPU time trying to
>> connect to a server that's OFF the network?
>>
>> Why are netscape windows so co-dependent? Sometimes one window gets
>> stuck, I can still use other programs, but not other netscape windows?
>>
>> Not that I feel like I'm in a position to demand anything from GPL
>> developers, just curious.
>>
>> Wroot
>>
>>
>> Sent via Deja.com http://www.deja.com/
2nd that Mozilla's a better bet - that last daily build's very good.
Running Galeon 8.3 on top of Mozilla is a resource saving step. I'm currently
testing the beta5 from Opera and so far so good. Lighter & pretty fast.
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Switching to Linux
Date: Fri, 5 Jan 2001 18:20:18 +0100
Lori Holder-Webb <[EMAIL PROTECTED]> wrote:
> Bash does fine by me, I'm not a power-scripter. Mostly I like the
> command line editing, the <TAB> command completion, the access to
> command line history, etc. Stuff I couldn't get from the csh
> implementation on the mainframe...
Then use tcsh.
>> In that case, I can understand it, because csh is to tcsh as sh is to
>> bash. You should use tcsh (it's STILL a much better shell than bash,
>> despite bashes years of playing at catch-up).
Peter
------------------------------
From: [EMAIL PROTECTED]
Subject: Adding a module when I've only got the binary
Date: Fri, 05 Jan 2001 17:43:23 GMT
The situation is, I've got a Dell server which has a RAID controller.
They only supply the driver in a binary format. Now I want to rebuild
the kernel. As you'd expect, there's no option in the kernel
configuration tools to include the driver. I tried the following:
- rebuilding the kernel & modules as per normal
- copy the driver into the relevant /lib/modules directory
- doing a depmod
I've tried various options on the depmod. First a depmod -a <version> ,
no good, then a depmod -a -F /usr/src/linux-<version>/System.map ,
still no good. I get unresolved symbols and, although a modules.dep is
created, when I try to boot I get unresolved symbols for the driver
and because none of the disks can be accessed, a system panic.
Anyone any ideas?
Thanks.
Tony Davis
Sent via Deja.com
http://www.deja.com/
------------------------------
From: "Chip Piller" <[EMAIL PROTECTED]>
Subject: Re: who's rewriting /etc/fstab?
Date: Fri, 5 Jan 2001 13:05:23 -0500
I am not sure what the problem is but it could be linuxconf. I was using
linuxconf on my RedHat 6.2 system and had a couple of system profiles.
Apparently linuxconf keeps an archive of configuration files for each
profile and this could be messing you up.
Is linuxconf running by default on your machine? If so turn it off (as
root, "/sbin/chkconfig linuxconf off" ) to see if your /etc/fstab still gets
changed when you reboot.
Hope this helps,
Chip
"Roger Davis" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
>
============================================================================
----
> I've encountered a bizarre problem in RedHat 6.2 where my /etc/fstab
> is being rewritten (incorrectly!) by either (i) PowerQuest's Drive
> Image Pro 4.0 backup/restore software or (ii) something which occurs
> during the RH 6.2 boot procedure. PowerQuest swears that their software
> does not do this, so I'm asking if anyone here is aware of any
> circumstances under which RH 6.2 would attempt to rewrite /etc/fstab
> on its own during boot-up.
>
> I am using PQ's Drive Image program to generate direct-to-CD-RW backups
> of the Linux ext2 partitions on my dual-boot (Win98/RH6.2) laptop.
> I wanted to verify the functionality of this backup, so I restored
> it to a spare partition on my hard disk. The restore procedure
> goes as follows: (i) run the PQ program from a DOS environment, using
> it to first delete the existing spare Linux partition and then immediately
> create a new Linux partition in the same spot on the disk and fill it
> with the restored filesystem read from the CD-RW, then (ii) exit the PQ
> program which causes a reboot. When the system attempts to reboot into
> RedHat 6.2 it fails because /etc/fstab has been rewritten incorrectly
> by agents unknown. This behavior is 100% reproducible.
>
> I immediately suspected the PQ program of attempting to be clever,
> looking around on my disk for my root filesystem's partition and then
> changing the fstab there to mount the filesystem that was just restored.
> PQ insists, however, that their program never writes anything whatsoever
> to any partition on the disk other than the partition to which the backup
> is being restored. Obviously their code has to alter the disk label
> to reflect the changed partitioning. I have no choice in the matter of
> deleting the existing partition and recreating it with the PQ restore
> software -- that's just the way their stuff works. The PQ backup image
> on the CD is supposedly a sector-by-sector copy of the hard drive
partition
> from which it was created. (Sectors not belonging to any file are not
> written to the image.) The new disk label appears to be correct, and after
> I muck around with fstab and reboot again the restored partition (and
> everything else) looks OK.
>
> The first three columns of my original /etc/fstab look like this (minus
the
> arrows at the right):
>
> /dev/hda5 / ext2
> /dev/hda2 /boot ext2
> /dev/hda6 /export/ctrobot ext2 <-------
> /dev/hda1 /export/ctrobotw98 vfat
> /dev/cdrom /mnt/cdrom iso9660
> /dev/hda7 swap swap <-------
> /dev/fd0 /mnt/floppy ext2
> none /proc proc
> none /dev/pts devpts
>
> Whoever is screwing up this file changes the first column of the two lines
> indicated by the arrows above to read
>
> /dev/hda5 /export/ctrobot ext2
> /dev/hda6 swap swap
>
> /export/ctrobot (which should be /dev/hda6) is the filesystem into which
> the data were restored, so it seems significant that it is the entry
> which gets screwed up, along with the adjacent swap partition (which
should
> be /dev/hda7).
>
> My partition table looks like this after the restore:
>
> Disk /dev/hda: 240 heads, 63 sectors, 3278 cylinders
> Units = cylinders of 15120 * 512 bytes
>
> Device Boot Start End Blocks Id System
> /dev/hda1 1 741 5601928+ 1b Hidden Win95 FAT32
> /dev/hda2 * 742 743 15120 83 Linux
> /dev/hda3 744 3278 19164600 f Win95 Ext'd (LBA)
> /dev/hda5 744 1021 2101648+ 83 Linux
> /dev/hda6 1022 1299 2101648+ 83 Linux
> /dev/hda7 3207 3278 544288+ 82 Linux swap
>
> Before the restore /dev/hda6 spanned cylinders 1022-3206, but it was
> downsized by the PQ restore program to be just large enough to hold
> the filesystem being restored (which was the backup of my root filesystem
> in /dev/hda5).
>
> Does anybody have any idea what the hell is going on here? Is there any
> way of examining the /etc/fstab on /dev/hda5 before I reboot into Linux
> with /dev/hda5 mounted as root so that I can verify whether PQ is in fact
> corrupting this file or not? (I tried a couple of years ago to make a
> rescue diskette containing a root filesystem and failed miserably after
> many hours of work. I am not eager to repeat that experiment! Is there
> a pre-built rescue diskette image containing a root filesystem that will
> boot my laptop available on the web anywhere?)
>
> Thanks much for any information that anyone can provide.
>
------------------------------
** 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
******************************