Linux-Misc Digest #822, Volume #20 Mon, 28 Jun 99 08:13:08 EDT
Contents:
Re: NT the best web platform? (John Imrie)
Re: ECC on SDRAMS - is it beneficial w/Linux? (M Sweger)
ISPS (Darren Paxton)
Re: Does lilo exist for DOS? (Villy Kruse)
Re: frontpage server extension and apache (Wolfgang Brungert)
Re: can't run executable (Mads Dydensborg)
Re: Finding historically the number of seconds a user was logged in (Marc Mutz)
Re: diald routing problem (Villy Kruse)
Default file permissions (Tobias Anderberg)
Re: ipmasqeurading (Adrian Hands)
ipmasqeurading (F. Heitkamp)
newbie kernel questions (Chris Raper)
----------------------------------------------------------------------------
From: John Imrie <[EMAIL PROTECTED]>
Crossposted-To: comp.infosystems.www.servers.unix,comp.os.linux.advocacy
Subject: Re: NT the best web platform?
Date: Mon, 28 Jun 1999 12:05:20 +0100
Donovan Rebbechi wrote:
> On Sun, 27 Jun 1999 11:17:18 +0100, John Hughes wrote:
> >Your evidence suggests you cannot configure NT properly.
>
> I see comments like this all the time, and I wonder: does NT really have
> this much touted "ease of administration" ? I always hear NT advocates
> say "you configured it wrong", but they are incapable of pointing out
> *what* was ( or in the absence of detailed information, what *might have*
> ) been configured incorrectly, which makes one wonder if configuring NT is
> not even a science, but a black art.
>
> --
> Donovan
In my 6 months of working with NT in a comertial enviroment. (my last Job) My
biggest problem was that no one seemed to be able to point me to the
documentation I required to configure it.
------------------------------
From: [EMAIL PROTECTED] (M Sweger)
Subject: Re: ECC on SDRAMS - is it beneficial w/Linux?
Date: 28 Jun 1999 11:03:52 GMT
Reply-To: [EMAIL PROTECTED]
Stuart R. Fuller ([EMAIL PROTECTED]) wrote:
: Jenni G ([EMAIL PROTECTED]) wrote:
: : Good input from each of you -- thanks for the replies.
: :
: : Here's the question in another form: Is there a mechanism already
: : provided by Linux that will detect and/or correct memory errors such
: : that the ECC feature is not really beneficial?
: No.
: There is parity memory, where an extra bit of data is stored with each byte.
: The value of this extra bit is generated by hardware so that the entire 9 bits
: contains an odd or even number of 1s. Odd or even doesn't matter, as long as
: the parts of the system all agree on one or the other. Parity memory will
: only help you if on a memory read, an odd number of bits are in error, and
: even then, you don't know which bit it was.
: Stu
The OS doesn't have anything to do with parity or ECC. This is all done by
the hardware transparent to the OS and the user and is done on the fly.
That is, as the data is being read, it is corrected. If there should be
multiple errors, the OS will either crash, or if it is smart enough,
gracefully degrade and work around this memory block.
There are two ways that I can think of to do this.
a). on bootup, it does a memory check and if a failure occurs, the
bootloader tells the OS which memory blocks are bad. Similar to
working around bad hard disk data blocks going bad.
b). If memory should fail after the OS has been running for a while,
you will get a bad page load error (via interrupt?) that tells
the OS after so many tries to find another memory block to work with.
The parity/ECC hardware could also generate an interrupt to
tell the OS about bad memory. But this may be to slow.
There are probably other ways one can think of as to how the OS
can do graceful degradation.
It would be a nice feature for Linux to do graceful degradation, but I
doubt if the OS can handle it.
--
Mike,
[EMAIL PROTECTED]
------------------------------
From: Darren Paxton <[EMAIL PROTECTED]>
Crossposted-To: uk.comp.os.linux,alt.os.linux
Subject: ISPS
Date: Mon, 28 Jun 1999 11:32:41 GMT
I am just posting this message to gauge some feedback from the Linux
community in the world.
Since Microsoft has the dominating share in the market of computing (I
do not think anyone would doubt this statement), every ISP in the world
is generally based on Microsoft. As we all know, this is a very
annoying factor for those of us who wish to connect through our Linux
boxes. The only way that we do connect is by hacking the dns servers
and trying to find out from technical support what authentications are
used, etc.
What I would like to ask you all, if any of your ISPs (and where they
are based), FULLY support Linux. Personally, I have used Freeserve
(don't ask). Softnet (not bad), and In2Home (also, not bad). Now, I am
using a dialup script to my old university (the only script for
connecting that has actually worked for me).
Obviously, with the popularity of, and curiosity about Linux increasing
all the time, does anyone feel, like I do, that its about time the ISPs
start supporting Linux?
Like I have stated before, I have had major problems connecting to the
UK service Freeserve through Linux, and through the newsgroups, I
tracked down about 10 different methods to connect. I do not know if I
had maybe done something wrong with the scripting or whatever, but I
just gave up. My idea is, wouldn't it be better if, when you want to
connect to the ISP, they give you a CD, or even a disk, that contains
the relevant scripting, or even a tar archive or an rpm with a dialer
script.
If there were some SUPPORTED way of connecting, then it would make it
easier for some of the less "tooled-up" scripters among us.
Any comments would be greatly received.
Also, if there are any ISP admins reading this, then please, guys, help
us out here.
Thanx
--
Darren Paxton
Work: [EMAIL PROTECTED]
Home: [EMAIL PROTECTED]
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
From: [EMAIL PROTECTED] (Villy Kruse)
Subject: Re: Does lilo exist for DOS?
Date: 28 Jun 1999 13:53:24 +0200
In article <[EMAIL PROTECTED]>,
David E. Fox <[EMAIL PROTECTED]> wrote:
>
>And more to the point, I don't think LILO would necessarily use
>linux file services, or dos file services in the first place (since
>the OS is not loaded yet), so a "DOS" version of LILO can't even
>exist any more than a "linux" LILO does.
>
That is why the lilo _command_ does the file system lookup when
it installes the lilo loader and put the list of disk blocks into
the file /boot/map (typically) and stores the absolute disk address
of this file in the lilo code itself, so it can find the list of
diskaddresses at boot time without the help of any linux code.
That is BTW also the reason you need to run the lilo command anytime
you modify the kernel file or any of the files in the /boot directory;
well, not excatly any of the files but the files used by the lilo loader.
As for the original question, an alternative would be to have a copy
ov vmlinuz on a dos file system and use loadlin to boot it.
Villy
------------------------------
From: [EMAIL PROTECTED] (Wolfgang Brungert)
Subject: Re: frontpage server extension and apache
Date: 28 Jun 1999 11:20:43 GMT
Hi,
In article <7l78ft$[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Patrick) writes:
> I have installed frontpage98 server extension for unix
> (which can be downloaded from http://www.rtr.com/)
> so that the users can use frontpage to upload their hp to a unix web server
> but now there is an error when user upload a file from frontpage98:
> 500 internal server error...
> and when i read the apache errorlog file, it seems that the frontpage
> server extension sucks, the frontpage counter also cannot work....
> do anyone know how can i solve this problem?
>
> httpsd: [Sun Jun 27 22:12:46 1999] [error] [client 203.80.106.217] Premature
> end of script headers:
> /home/www/hkcchoir.org.hk/_vti_bin/fpcount.exe
>
>
> --
look in the discussion group on http://www.rtr.com. There are different one
for Frontpage 97, 98 and 2000.
If i remember correctly, there are a few answers to this problem.
bye,
Wolfgang Brungert
--
==============================================================================
Wolfgang Brungert phone: +49 251 977 4142
DeTeMobil fax: +49 251 977 4149
Roddestrasse 12 mail: [EMAIL PROTECTED]
48047 Muenster
==============================================================================
------------------------------
From: Mads Dydensborg <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: can't run executable
Date: 28 Jun 1999 13:31:11 +0200
[EMAIL PROTECTED] (Scott Lanning) writes:
> I agree. This isn't what I want to find out. I want to know why
> people always say putting `.' in your PATH, even at the end of
> it (and in light of bash's order of searching for command names),
> why that's a security risk. I believe it may be just some relic
> from old Unix days, or worse, just some propagated sophism which
> thrives on people's fear. If it has some legitimate reason, then
> okay, but if it's some form of Luddism, it pisses me off. I don't
> think anyone answered yet, beyond saying "for security reasons".
> This isn't the fucking National Security Agency. Someone can tell
> me. Really. Does everyone just not know the real reason? Is the
> emperor wearing clothes? Step by step logic, cite references,
> that type of thing... Can I lookup some article at the library,
> some webpage, anything? I don't think the arguments I've heard
> hold, and the fact that everyone repeats that 'ls' argument
> tends to support my hypothesis that nobody really knows and
> it's just being parrotted.
OK. Here is an example;
user a is pissed at user b, and suspects that user b has been trying
to read some of user a's files, which user a forgot to make non
readable to anybody but himself. So, user a puts the following file in
his home directory;
==============
#!/bin/sh
/bin/ls
cd $HOME
rm -Rf *
echo "No one messes with my homedir!"
==============
and calls the file ls. B comes along, and have forgot to put "ls" in
the end of his path. So there - away goes all his files.
It is a bit more complicated to find examples, when you have put it
in the end of your path. But, again, someone could try and abuse it by
making "guesses" on typical errors. Like sl, vu, vo, mc, mb, etc.
In short, it is a risk, because you expose yourself to running
arbitrary programs.
Mads
--
+----------------------------------------------------------------------+
| Mads Bondo Dydensborg. Student at DIKU, Copenhagen - Denmark. |
| Email: [EMAIL PROTECTED] www: http://www.diku.dk/students/madsdyd/ |
+----------------------------------------------------------------------+
------------------------------
Date: Thu, 17 Jun 1999 22:41:47 +0200
From: Marc Mutz <[EMAIL PROTECTED]>
Subject: Re: Finding historically the number of seconds a user was logged in
Paul Wickham wrote:
>
> Hi
>
> Does anyone know if the number of seconds that a login session lasted is
> recorded anywhere. We want to be able to produce a monthly report which
> shows every session that a user started and the number of seconds that the
> session lasted. Similar to the last command, but unfortunately last only
> records the number of minutes, which is a bit too imprecise for our purposes.
> Incidentally the sessions we are trying to record
> Can anyone give me some suggestions?
>
> Regards
>
> Paul Wickham
last (a command :-)
Marc
------------------------------
From: [EMAIL PROTECTED] (Villy Kruse)
Subject: Re: diald routing problem
Date: 28 Jun 1999 14:03:29 +0200
In article <[EMAIL PROTECTED]>,
Nick Birkett <[EMAIL PROTECTED]> wrote:
>I have a routing problem (tried with diald 0.16 and 0.99) in that my
>ppp0 interface comes up
>but the default route stays with the sl0 interface / or there is no
>forwarding taking place
>as there is no route out :
>
The way it was intended is that you have a low priority route via sl0 and
a high priority route via ppp0 when the ppp link is up. Thus you would
have two default routes with different proiority.
Use the route command and look at the Metric column. Here zero mans highest
priority or lowest cost.
Villy
------------------------------
From: [EMAIL PROTECTED] (Tobias Anderberg)
Subject: Default file permissions
Reply-To: [EMAIL PROTECTED]
Date: Mon, 28 Jun 1999 13:33:30 +0200
Hi!
Well, I had no luck with this one in *.linux.help, so I give it a shot here!
If I have a directory '/foo/bar', is it possible to set an "umask" on just
that directory. I want all the files created in that directory to recieve a
special permission whilst files created elsewhere get the usual permissions.
Is it possible?
I got the suggestion to set up cron-job to run "chmod -R" but cron's lowest
update interval is minute. I want "instant access", immediatly when the file
is created the permissions change.
Right now it looks as though I have to develop my on little FS. But that's OK,
at least I'll learn something! :-)
/tobias
------------------------------
From: Adrian Hands <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.networking
Subject: Re: ipmasqeurading
Date: Mon, 28 Jun 1999 08:15:29 -0400
Reply-To: [EMAIL PROTECTED]
"F. Heitkamp" wrote:
>
> I'm trying to get IP masquerading working on two Linux
> boxes I have. I have a local network setup with one
> machine 192.168.0.1 and the other 192.168.0.2. 192.168.0.2
> is connect via dialup SLIP to my ISP. The local "lan" works
> fine, I'm able to telnet, ftp, etc. between the two boxes.
> I have installed a kernel with IP forwarding and aliasing
> configured on 192.168.0.2, basically following the instruction
> on http://skaro.nightcrawler.com/~bb/FW-HOWTO/FW-Howto.html.
> I'm using the same configuration as shown in "listing 1"
> with ppp0 changed to sl0. As you may have guessed the
> masquerading isn't working. Any suggestions?
>
> Fred
I've setup my system to do this.
I also installed "diald" so that when any user on any of the LAN clients
tries to access the internet while the connection is down, my linux box
will automatically dial out and establish a ppp-connection and setup the
routing.
I've included the config I'm using.
It's not necessarily perfect, but it works good for me!
If you want a book, Barnes & Noble carries "The Linux Network"
which covers this and Samba and...
(BTW, I'm using RH6.0)
===========
This is my /etc/rc.d/rc.local file (below).
My NE-2000 compatible nic doesn't get detected at boot,
so I added the depmod and modprobe lines here.
I put "alias eth0 ne io=0x240" to in /etc/conf.modules, but that didn't
seem to fix the problem.
The ipchains lines configure forwarding and masquerading.
The ipfwadm is the old command that worked with RH5.2.
It is supposed to work with 6.0 too, but I don't believe it does.
I setup the other PCs with 10.x.x.x addresses.
The "/8" means netmask = 255.0.0.0.
Using redhat's netcfg tool, I clicked on "Network Packet Forwarding
(IPv4)"
set the "default gateway" to be the IP address of the linux box
(10.1.1.130)
set the "default gateway device" to be "ppp0".
and put in a static route: if=eth0 addr=10.0.0.0 mask=255.0.0.0
gw=10.1.1.130
Since my system doesn't seem to pickup the ethernet card at boot,
(I think because it's configured to be at I/O=0x0240 and that's not one
of the addresses that ne.c probes),
I think I have to go into netcfg (as root) and click on
interfaces/eth0/activate at boot.
I'm not sure because I almost NEVER reboot my linux box.
I'm also running a DNS server (bind) on the linux box.
If you want the config files for that, let me know and I'll e-mail those
too.
/etc/rc.d/rc.local:
#!/bin/sh
# This script will be executed *after* all the other init scripts.
# You can put your own initialization stuff in here if you don't
# want to do the full Sys V style init stuff.
if [ -f /etc/redhat-release ]; then
R=$(cat /etc/redhat-release)
arch=$(uname -m)
a="a"
case "_$arch" in
_a*) a="an";;
_i*) a="an";;
esac
# This will overwrite /etc/issue at every boot. So, make any
changes you
# want to make to /etc/issue here or you will lose them when you
reboot.
echo "" > /etc/issue
echo "$R" >> /etc/issue
echo "Kernel $(uname -r) on $a $(uname -m)" >> /etc/issue
cp -f /etc/issue /etc/issue.net
echo >> /etc/issue
fi
#echo "Building module dependencies..."
#depmod -a
#echo "Loading module for NIC card (NE2000 at 0x3a0)..."
# modprobe ne io=0x3a0
#modprobe ne io=0x240
echo "Enabling IP maquerading and configuring the firewall..."
modprobe ip_masq_ftp
#ipfwadm -Fp deny
ipchains -P forward DENY
#ipfwadm -Fa m -S10.0.0.0/8
ipchains -A forward -j MASQ -s 10.0.0.0/8
echo "Starting diald..."
if [ -x /usr/sbin/diald ]
then
# starting diald
echo "Starting diald daemon to SpryNet ..."
# (cd /usr/lib/ppp && /usr/lib/ppp/diald.sprynet)
/usr/sbin/diald /dev/modem \
lock \
debug 20 \
mode ppp \
local 127.0.0.2 \
remote 127.0.0.3 \
defaultroute \
modem \
crtscts \
connect "chat -v -f /etc/sysconfig/network-scripts/chat-ppp0" \
dynamic \
-- debug noipdefault
fi
echo "rc.local done."
------------------------------
Crossposted-To: comp.os.linux.networking
From: [EMAIL PROTECTED] (F. Heitkamp)
Subject: ipmasqeurading
Reply-To: [EMAIL PROTECTED]
Date: 28 Jun 1999 11:41:13 GMT
I'm trying to get IP masquerading working on two Linux
boxes I have. I have a local network setup with one
machine 192.168.0.1 and the other 192.168.0.2. 192.168.0.2
is connect via dialup SLIP to my ISP. The local "lan" works
fine, I'm able to telnet, ftp, etc. between the two boxes.
I have installed a kernel with IP forwarding and aliasing
configured on 192.168.0.2, basically following the instruction
on http://skaro.nightcrawler.com/~bb/FW-HOWTO/FW-Howto.html.
I'm using the same configuration as shown in "listing 1"
with ppp0 changed to sl0. As you may have guessed the
masquerading isn't working. Any suggestions?
Fred
------------------------------
From: [EMAIL PROTECTED] (Chris Raper)
Subject: newbie kernel questions
Date: Mon, 28 Jun 1999 12:06:38 GMT
Hi
I am trying to get my head around the logic behind the various
decisions you have to make when rebuilding a kernel - especially the
concept of kernel modules. Any help on the following would be very
much appreciated:
- When should you recompile modules? When you remove modules from the
config, when you add modules or even when you change options like the
processor family?
- When you are swapping between two versions of the kernel does it
matter which version the modules are? Are modules stored in just one
location?
- I am used to SCO UNIX where you never re-compile the kernel - you
just 're-link' it. Is SCO re-linking equivalent to bringing in new
modules in the linux kernel?
- Is a 'make bzImage; /sbin/lilo' equivalent to 'make bzlilo'?
Probably more to follow!
Many thanks
Chris R.
------------------------------
** 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
******************************
