Linux-Setup Digest #30, Volume #21 Wed, 11 Apr 01 14:13:18 EDT
Contents:
Re: Red Hat system start up (H.Bruijn)
Re: Can't print (drivers?) ("Peter T. Breuer")
Setting up form defs for linux printing. ("jf")
Re: Executable vmlinux in /boot ("Peter T. Breuer")
Re: how use map command in vim?? (H.Bruijn)
Re: Puzzling Lilo problem with backup and restore via tarball on CD-R ("Peter T.
Breuer")
Re: Where to find mkinitrd-2.9.1.src.rpm? (Tim J)
Re: CVS on RH7 ("Sandro Leopardi")
burning cds and the like...... (Michael Falconer)
Re: can't set up remote filesystems with SuSe 7.0 (Brian O'Halloran)
Re: make config (The Ghost In The Machine)
Partitioning strategy for multiple Linux distributions? ("Richard Scott")
General setup guide lines (Mohamed Sentissi)
Xconfigurator help. ("Kenny@BUI")
Want to dual boot with win 2000. ("blkthorn")
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (H.Bruijn)
Subject: Re: Red Hat system start up
Date: 11 Apr 2001 14:14:03 GMT
Reply-To: [EMAIL PROTECTED]
On Wed, 11 Apr 2001 01:18:56 -0700, Craig Van Tassle allegedly wrote:
>I am trying to figure out how to change the way my system starts up so i
>can close down several services that I don't want to be open (netbios
>for one) I was wondering if any one know a t-phile that explains the RH
>start-up system in detail. the only ones that I found said it uses
>scripts. Well what do i need to do to modify those scripts to kill the
>things that I don't want open
Nearly all linux distibutions use the SysV system to control the boot
and start-up sequence, Slackware is IIRC the only one that uses BSD
style startup scripts.
There are six runlevels, each with a different meaning:
0 - halt; all services are killed, disks unmounted and the systems shuts
itself down.
1 - Single user mode; no network, only login from console, used to do
major system maintenance.
2 - Multiuser, without NFS (The same as 3, if you do not have networking)
3 - Full multiuser mode; login from multiple consoles, via the network.
All daemons and services are running and accesible.
4 - unused
5 - X11; The same as 3, but instead of booting to a console login prompt,
X11 is started and you get a Graphical User Interface.
6 - reboot all services are killed, disks unmounted and the systems
reboots.
The boot sequence is as follows. First the operating system is selected
with lilo. Depending on your situation you select either a different OS,
fi windows or OS/2 or a linux kernel version. These options are set by
updating the /etc/lilo.conf ("man lilo.conf") file followed by running
the command "/sbin/lilo -v" ("man lilo").
Assuming that now you have selected linux, the linux kernel is loaded.
It generally gives output as it detects your hardware, it finds network
adapters, ide controllers, scsi controllers and more. The capabilities
of the kernel are partly hardcoded into the kernel, and can partly be
extended by loading kernel-modules. Kernel-modules are "drivers" that
can be loaded by the running kernel, by linux, as needed. They can be
unloaded again when not used anymore. This allows the kernel to be small
and compact, and useable on different configurations by loading only
those modules needed on that configuration. The properties of the
modules are configured with the file /etc/modules.conf ("man
modules.conf") and can be updated on a running system with the command
"/sin/depmod -av" ("man depmod"). To change the properties of the kernel,
it generally needs to be recompiled. Download the source, change the
config, compile and install the kernel.
After loading the kernel you get the second stage of the startup
sequence, init ("man init"). This part is controlled by the file
/etc/inittab ("man inittab"). What happens depends a lot on the
configuration file, and possibly arguments given when you selected linux
at the lilo boot prompt, but generally there is first a general system
configuration, followed by entering the default runlevel.
The general system configuration is mostly compromised of: cheching the
integrity of filesystems, mounting them, setting the keymaps,
configuring hardware, isa cards, the serial port etc. it sets the
hostname and more. On RedHat systems this is controlled with the file
/etc/rc.d/rc.sysinit on Debian in the scripts contained in /etc/rc.boot
and /etc/rcS.d/.
Then the default runlevel is entered. What happens there is that for all
daemons started (or stopped) and other tasks that need to be executed
there is a script in the directory corresponding to that runlevel. FI
for runlevel 3 the scripts will be in the directory /etc/rc.d/rc3.d/ (on
RedHat, on Debian systems it would be /etc/rc3.d/ and SuSe likes to keep
those in /sbin/ somewhere). Now if you looh closely you'll notice that
those scripts aren't stored in that directory, but are in fact symolic
links to scripts in /etc/rc.d/init.d (or /etc/init.d). This saves space,
now you only have one file for each daemon or task, instead copies of it
in each runlevel, and it makes maintenance easier as well, with only one
file to edit, instead of several.
Anyway in the directory /etc/rc.d/rc3.d/ are the scripts (symlinks)
which are executed upon entering. When you look you'll notice an odd
naming convention, all names start with either a capitol K or S,
followed by a number between 01 and 99, followed by a name of a daemon or
task. The scripts first executed are the ones whose names start with a
K, in the order they are numbered, then by alphabet. Those are executed
with the option "kill", which results in the daemon being shut down, if
it was up and running. The ones with an S will be executed with the
option "start", and thus starting a daemon or service, if it wasn't
already. The precise results depend on the script.
As you can imagine entering runlevel 3 will have only one script with a
K, only X11 will be closed when entering runlevel 3 from runlevel 5,
everything else remains running, or gets started.
Shutting the system down, in other words changing runlevel to 0, of
course means that all programs are closed, the users get kicked off (so
always be nice and give notice to your users with either "wall" or use
the command "shutdown -t 600 "System will go down for maintenance at
exactly 15:00h. Please close all your programs, save your data and log
out."") So rc0.d/ will have nearly all symlinks start with a K, except
for the one that does the actual powerdown.
So to change what gets started at a runlevel, go to the directory
associated with that runlevel, and move the symlink with K to S and the
otherway around. The numbering isn't completely trivial, so don't just
pick one, ut place the script correctly in the order. For example don't
start any daemons before the system logger syslogd has been started.
When you use your own nameserver, make sure it starts before you start
any of the other network related stuff, they often depend on resolving
the names of other servers they need to contact, or resolving the
adresses requests originate from. Etc. Use your common sense.
--
If a trainstation is the place where trains stop, what is a workstation?
========================================================================
Herman Bruijn website: http://hermanbruijn.com
The Netherlands
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.redhat
Subject: Re: Can't print (drivers?)
Date: Wed, 11 Apr 2001 16:07:27 +0200
Bill Unruh <[EMAIL PROTECTED]> wrote:
> In <[EMAIL PROTECTED]> "Peter T. Breuer" <[EMAIL PROTECTED]> writes:
> ]Taavi Hein <[EMAIL PROTECTED]> wrote:
> ]> : Absolutely none. You don't "get drivers" like you do with M$ Windows.
> ]> : Tell PrintTool that it is an HP DeskJet 550.
> ]> : You could have found this information at:
> ]> : http://www.linuxprinting.org/
> ]> Are you telling me, Linux doesn't use drivers (or anything of a similar
> ]Not for printers, no ... they were originally one-way devices. You
> ]don't need anything apart from the standard byte-blatting printer driver
> ]to "drive" them. Just make sure to send them the right bytes, which
> ]you'll get by transforming your input via a filter ...
> Well, not really true. Printers have a huge bunch of different commands
It's very true. The commands (the language spoken) is irrelvant, since
no reply is expected or necessary. Just convert the text to that
language offline (using a filter) and throw it at the generic printer
driver, which will do the _driving_ necessary.
> to do things and the printer driver converts the output to what th
No it doesn't. That's what a FILTER does.
> eprinter wants. For example, you can set your printer up a postscript
> and then use ghostscript to convert postscript to what the printer
> wants.
Is there some reason that I carefully explain the difference between a
filter and a driver, and the reason that you don't need a driver, just
a filter, and then you go and disregard the entire carefully worded
post ...
> ]> Well, I don't really care, if PrintTool uses 'drivers' or 'filters' or
> ]You should ... they're very different animals.
> well, not really.
Really.
> ]Drivers start being necessary when we need to know more about the state
> ]of the printer in order to be able to "drive" it effectively. But a
> ]printer really only needs to be able to say "hold on a bit", or
> ]"error", or "ok". so it's no surprise that the generic driver will do
> ]for most. It's when the printer is bereft of reason that one needs to
> ]start writing the driver that its creators forgot to embed in it.
> Not true. Try sending HP Laserject graphics commands to a Bubblejet
> printer. Or postscript to a dot matrix printer.
That is not the business of a driver. A driver handles HARDWARE level
states and interrupts. You are talking about a COMPILER (also called a
filter or translator) which is a single input, single output animal whose
job it is to change the format of texts. No state machine involved, cough,
cough .. you can do the filters conversion offline. That should tell
you the difference.
> ]> : Best advice, buy a hardware modem.
> ]> Not the best (from my point of view), but the easyest one ;) If I had the
> ]> money, I'd already have a real modem, instead of searching for help.
> Buy a real modem. Find a neighbour whose grass you can cut to get the
> money. It will take far far less time than the time you waste trying to
> get a winmodem (even if drivers exist) to work and then to do the whole
Well, if he's really in the situation of having to mow the lawn for
money, it might well be useful to compile the ltmodem driver just so he
can earn money by doing it for other people too!
> thing again each time you upgrade the kernel or distribution.
Peter
------------------------------
From: "jf" <[EMAIL PROTECTED]>
Subject: Setting up form defs for linux printing.
Date: Wed, 11 Apr 2001 09:30:19 -0500
Hello!
We're moving away from SVR in favor of Linux (Red Hat Linux 6.2 Enterprise
Edition) Kernel 2.2.14-11, but I'm stuck! In UNIX, there's an 'lpforms'
command utility that helps configure print forms for various printers. I
can't find a similar utility in Linux...is there one? (or is this one of the
'bells and whistles' not available in Linux).
Thanks in advance,
JF
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Executable vmlinux in /boot
Date: Wed, 11 Apr 2001 16:11:18 +0200
Mladen Gavrilovic <[EMAIL PROTECTED]> wrote:
>> Redhat has for the past few releases installed the uncompressed vmlinux
>> file into the /boot directory, and gives no reason I could find for that.
>>
>> It is remotely possible that some commands will search the symbol table
>> of vmlinux rather than reading the /boot/System.map for symbol values,
>> as is quite common on traditional unix systems.
> Thanks, that's the answer I've been looking for.
Except it's not an "answer", but as the author says, "a remote
possibility". Neither of us can think of any actual reason for them
to do so. It's more likely to be an error on their part than anything
else we can propose. System.map is complete.
Another remote possibility is that they've appended the .config to the
vmlinux image? No, no way.
Peter
------------------------------
From: [EMAIL PROTECTED] (H.Bruijn)
Subject: Re: how use map command in vim??
Date: 11 Apr 2001 14:38:39 GMT
Reply-To: [EMAIL PROTECTED]
On Wed, 11 Apr 2001 20:56:31 +0800, hushui allegedly wrote:
>Can somebody give me a example?
>:map ^i include
>Is it ???
>But it does not run.
>I means define a shortcut key of mine in vim so that I can input my program
>quickly.
>
I would suggest asking this in news:comp.editors or at one of the lists
suggested at the site http://www.vim.org or search that site.
--
If a trainstation is the place where trains stop, what is a workstation?
========================================================================
Herman Bruijn website: http://hermanbruijn.com
The Netherlands
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Puzzling Lilo problem with backup and restore via tarball on CD-R
Date: Wed, 11 Apr 2001 16:29:26 +0200
Lucius Chiaraviglio <[EMAIL PROTECTED]> wrote:
> It ALMOST works right. The problems are that in some cases I end up with
> Lilo hanging at "LI" when the new disk is installed on its target system,
> and that even when this does not happen, attempting to run the lilo command
> on the new disk fails with the error message "First sector of /dev/hda1
> doesn't have a valid boot signature." I could not find any reference to
That means you said it's an msdos system and included it in lilo as
such, but it isn't (at least not one lilo can do anything with!).
> such an error message nor (more importantly) how to fix it in the man or
Look in the code.
> info pages for the lilo command, nor in any of the information about Lilo
> (and not Lilo + some non-Linux operating system) on http://www.linuxdoc.org.
> I did a grep for this error message in the Lilo source directory from a Red
> Hat 6.2 source CD, and found it and the BOOT_SIGNATURE constant 0xAA55, and
> also found this constant (with the name SIG_1) in
> /usr/src/linux-2.2.14/arch/i386/boot/setup.S. This showed me what is
> missing, but gave no indication of how to fix it, other than (possibly) by
> using dd and "vi -b" (or the equivalent*) to hack the first sector of
> /dev/hda1 to fix the appropriate bytes.
Well done! Yah, go for it! What can happen ?;-).
> (* Anyone want to recommend a hex editor, especially if it can work directly
> on disk sectors, or really especially if it can act as a Linux-based
> substitute for Norton Disk Editor? I could really use this for tackling
> another problem.)
I use bvi. I presume hexedit does the same thing.
Peter
------------------------------
From: Tim J <[EMAIL PROTECTED]>
Subject: Re: Where to find mkinitrd-2.9.1.src.rpm?
Date: Wed, 11 Apr 2001 14:42:03 GMT
Thanks, that's a good site to know.
tech2kjason wrote:
> I remember upgrading the mkinitrd binaries to mkinitrd-3.0.10-1,
> without any problems, and though I download the same 3.0.10-1 source
> rpms, I never had to put them to any use... If nothing else, you may
> be able the 2.9.1.src.rpm
>
> http://www.rpmfind.net
>
> On Tue, 10 Apr 2001 20:58:45 GMT, Tim J <[EMAIL PROTECTED]> wrote:
>
> >I am in the process of upgrading my rh7.0 kernel to v2.4. In the
> >Changes document it suggests upgrading to version 2.9 and that I can
> >download it at.
> >ftp://rawhide.redhat.com/pub/rawhide/SRPMS/SRPMS/mkinitrd-2.9.1.src.rpm
> >
> >This directory could not be found. I looked around Redhat's site and
> >could not find this rpm.
> >Where can I find it and if it is not available do I need it to compile
> >kernel 2.4?
> >
> >
------------------------------
From: "Sandro Leopardi" <[EMAIL PROTECTED]>
Subject: Re: CVS on RH7
Date: Wed, 11 Apr 2001 16:19:09 +0200
the way to translate your message into xinetd is to write a file named
cvspserver into the /etc/xinetd.d/ directory.
file contents should be like this:
service cvspserver
{
socket_type = stream
protocol = tcp
wait = no
user = root
server = /usr/local/bin/cvs
server_args = --allow-root=/usr/local/newrepos pserver
disable = no
}
hope this helps...
you should restart xinetd after you write the file in order for the service
to be active
Sandro
"maher" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED]...
> hi there,
>
> I got a problem to configure CVS under RH7.Seems that RH7 use xinetd
> whereas the previous/older redhat use /etc/inetd.conf .Quite confusing
> to translate these messages into xinetd :
>
> In the /etc/inetd.conf,I need to add this
>
> cvspserver stream tcp nowait root /usr/local/bin/cvs cvs \
> --allow-root=/usr/local/newrepos pserver
>
> But in xinetd ? How do i translate them into xinetd format?
>
> need your help and expertise :)
>
> thanks in advance
>
> maher
------------------------------
Date: Wed, 11 Apr 2001 17:28:46 +0100
From: Michael Falconer <[EMAIL PROTECTED]>
Subject: burning cds and the like......
Hello there. I have successfully managed to recompile the kernel,
2.4.1 to incorporate the scsi emulation etc that is needed to access
a ide cdrom burner. (HP8100+). Using Mandrake 7.2. All is fine. I can
burn data cds,I can burn audio cds. I can even mount cds that I was
using for a giant floppy under Windows. However I cannot burn to the
cd that I am using as a giant floppy.
Every time I try and mount the cd , it mounts it ok, but
will only mount it in ro mode. It will not mount in rw mode. I have
read a lot of docs about this, including of course, mount and also
mtab and fstab. But to no avail. Is this yet possible under Linux.
Hope someone out there has the answer!!! Thanx in advance.
Mike
--
I counsel thee to buy of me gold tried in the fire, that thou mayest
be rich; and white raiment, that thou mayest be clothed, and that the
shame of thy nakedness do not appear; and anoint thine eyes with
eyesalve, that thou mayest see. ( Rev 3:18 AV)
------------------------------
From: Brian O'Halloran <[EMAIL PROTECTED]>
Subject: Re: can't set up remote filesystems with SuSe 7.0
Date: Wed, 11 Apr 2001 17:43:14 +0100
Hi all.
Just a quick note to all who replied with advice regarding mounting
remote filesystems on a machine running SuSe 7.0. Thankfully the problem
is now fixed - as it turned out, it was a problem regarding permissions
on the /etc/mtab file. Always the simple things eh!
Thanks again to all.
Brian O'Halloran
------------------------------
From: [EMAIL PROTECTED] (The Ghost In The Machine)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: make config
Date: Wed, 11 Apr 2001 17:20:20 GMT
In comp.os.linux.advocacy, ja
<[EMAIL PROTECTED]>
wrote
on Wed, 11 Apr 2001 21:57:25 +0800
<9b1nsj$188m$[EMAIL PROTECTED]>:
>my system redhat7.0
>kernel 2.2.16
>
>I extract pcmcia-cs-3.1.15 at /usr/src/linux
>make config
>why can I input "/usr/src/linux" as my linux source directory?
>
>configuration failed appear~
>
>why?????????
Try asking in comp.os.linux.setup; this is an advocacy group.
Followups directed thereto.
(Is there a comp.os.linux.kernel.setup?)
I will also note that "pcmcia-cs-3.1.15" could be anything;
is it:
- a tarball (.tar.gz or .tgz)?
- Redhat or Mandrake package (.rpm)?
- Debian package (.deb)?
- something else with which I'm not familiar, such as SuSE, Phat,
or Monkey?
Also, the usual permissions and kernel rebuild issues (were you root?
is /usr/src/linux open? did you rerun 'make config'?) apply.
Good luck. I don't have an answer, unfortunately.
>
>
>[EMAIL PROTECTED]
>
>
--
[EMAIL PROTECTED] -- insert random misquote here
EAC code #191 5d:19h:08m actually running Linux.
Linux. The choice of a GNU generation.
------------------------------
From: "Richard Scott" <[EMAIL PROTECTED]>
Subject: Partitioning strategy for multiple Linux distributions?
Date: Wed, 11 Apr 2001 12:39:11 -0500
I have a 30G IDE hard drive and I'd like to install up to
four Linux versions for comparison.
What is the best partitioning strategy to do this?
I'd like to install LILO in the MBR.
Richard
------------------------------
From: Mohamed Sentissi <[EMAIL PROTECTED]>
Subject: General setup guide lines
Date: Wed, 11 Apr 2001 11:35:27 -0600
hello
I have to write a guide line for some people who gonna be making linux
client boxes using RH7.
I have to make a guide line on what the packages they should select to
install and what not. I know a little bit on what I should advise but
please can smbdy tell more if they have a suggestion ?
please ?
like what not to install because it's a security hole since those linux
clients gonna be on a campus network ?
thanks
------------------------------
From: "Kenny@BUI" <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.misc
Subject: Xconfigurator help.
Date: Wed, 11 Apr 2001 14:06:41 -0400
hello,
we just upgraded to from 6.2 to 7.0.
the thing would not display properly. we are trying to use Xconfigurator to
fix the problem. it appears that the monitor is being detected but when we
type startx the
screen does not refresh properly. the icons are there but you have to put
the mouse over them for them to appear. there are also lines all over the
screen. looks like the tile effect in win98.
------------------------------
From: "blkthorn" <[EMAIL PROTECTED]>
Subject: Want to dual boot with win 2000.
Date: Wed, 11 Apr 2001 18:07:04 GMT
I installed Win2000 and want to dual boot with Mandrake 7.1. Here what I
got so far.I fdisk my drives with fdisk and format both drives one is c:\
and other is d:\ thought Win2000, now should i go back in fdisk and get rid
of d:\ drive and let Mandrake do the partition or do I just install
Mandrake and it will istall it's self on drive d:\.
thanks for your help: blkthorn
------------------------------
** 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.setup.
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-Setup Digest
******************************