Next Up Previous Contents Index
System Administration

E.9 System Administration

E.9.1 When I create users with usercfg, they can't login.

If when using usercfg the password is never encrypted when typed in or only one asterisk is placed in the password field, it is a usercfg bug. We are working to fix this bug, but currently we have this work around.

Click the button to the left of the entry field and choose change, then type the password TWICE in the entry field (follow it with a return each time).

Another alternative is to set their password using the command line passwd program.

E.9.2 What is PAM? Why use it?

PAM is a standard adopted by other unices such as Solaris 2.6. For more information on PAM please read: /urlhttp://www.redhat.com/linux-info/pam/

E.9.3 How can I setup Secure Shell (SSH) on my linux system?

Due to United States of America export restrictions on munitions, Red Hat Linux can not distribute ssh. The site ftp.replay.com has set up various downloads of ssh and PGP.

E.9.4 Why does Linux only see part of my RAM?

There are a couple of things that could be causing Linux to not see all your memory. On some 386's you need to compile your kernel with `Limit memory to 16M?" enabled.

On most systems, the reason is that the BIOS has a limit of how much memory it will tell the OS is present in the machine, even though the board can have more. Common limits seen with this problem are 16M, 32M, 64M, and 128M. To get around this, we need to explicitly specify the amount of memory to the kernel at boot time via the mem=< actual memory goes here > flag.

In the following example, we have a 32M machine but only 16M are being seen by Linux. At the LILO prompt, we type

LILO: linux mem=32M

After the machine boots, we use the free command to see if the larger amount of memory was recognized by the kernel. If so, we can add an append line to the /etc/lilo.conf file and rerun lilo to make it happen permanently. The example from above could look like the following:

boot=/dev/sda
map=/boot/map
install=/boot/boot.b
prompt
timeout=50
image=/boot/vmlinuz-2.0.32
        label=linux
        root=/dev/sda1
        initrd=/boot/initrd-2.0.32.img
        read-only
        append="mem=32M"

Do not forget to run lilo after editing the file.

E.9.5 I have over 64 Megs in my pentium or greater machine, however it seems sluggish when doing anything. If I tell Linux to use only 64Megs with the mem=64M machine, it seems to speed up. What is going on and can I do anything about it?

This problem is a problem with hit rates and memory. A few motherboards do not use the external cache at all if the machine tries to access memory more than 64M in size. This will cause a noticeable slow down, and there is not much that can be done about it. Most other problems is that the CPU is not using the External Cache if you only have 256K cache and more than 64Megs of ram. Increase the Motherboard cache to 512K to 1Megabyte and you may see the sluggishness go away.

E.9.6 I have Red Hat 5.0 and have upgraded to the ld.so RPM package listed in the errata, but my libc5 applications still seg fault. What is wrong?

The problem with crashing libc5 applications can be due to several items.

E.9.7 When I run fstool, I get a message that says a partition seems to have been deleted and asks if I want to remove it from /etc/fstab.

The fstool program is not working properly with current versions of tcl and shouldn't be used. It should have been obsoleted, but slipped through the cracks.

First we will have to fix the /etc/fstab file since fstool may have corrupted it. The areas that seem to be changed by fstool are usually the cdrom and swap. Here are sample lines (you will need to change the partitions to match your system.)

/dev/sda2            swap                swap  defaults      0 0
/dev/cdrom           /mnt/cdrom          iso9660 noauto,ro   0 0

You should now remove the fstool program using rpm.

rpm -e fstool

Please use cabaret which is a functional working alternative. This program is located in /usr/sbin/cabaret and can be executed from the command line.

E.9.8 How do I configure my Jaz drive under linux?

Documentation on using Jaz with Linux can be found on the cdrom in
doc/HOWTO/mini/Jaz-Drive and on the system in /usr/doc/HOWTO/mini/Jaz-Drive.

E.9.9 How do I use my parallel port zip drive?

Here's something you can try: edit /etc/conf.modules and add the following line to the others

alias scsi_hostadapter ppa

If you need to send the ppa driver any options about which LP is being used etc, you would add the line

options ppa ppa=<options go here.>

For more information, check www.torque.net/paraport

E.9.10 I'm having problems getting my IDE zip drive to work.

First check and make sure there is a disk in the drive. Also, make sure you are mounting it as partition 4 instead of 1. An example would be hdc4. The supposed reason for it being partition 4 is that Mac SCSI uses this partition as it primary partition.

E.9.11 Some of my older applications get the incorrect time.

Some libc5 apps want /usr/lib/zoneinfo, so you can either recompile them for libc6 or provide a symlink so that things will work.

        ln -s ../share/zoneinfo /usr/lib/zoneinfo

Please see also check the Red Hat errata (www.redhat.com/errata) for other items.

E.9.12 I have all the latest updates installed, but my programs still get the incorrect time.

If you have installed all the latest updates and you programs still get the incorrect time, try checking the settings in /etc/sysconfig/clock. They probably look something like this:

UTC=true
ARC=false

This means that Linux will assume that your BIOS clock is set to the UTC or GMT timezone. More than likely, the clock is set to your local timezone, and you need to change the UTC line to be

UTC=false

E.9.13 During the install, I was not asked to use the 2nd cdrom. When I use the X program glint on it, it reports that there are no rpms, but when I look at the directories, I see lots of them. What is going on?

The 2nd cdrom in the Red Hat Linux boxed set contains the source code rpms (SRPM) for all of the Open Source applications that are on the first cdrom. From these source rpms, you can build all the Open Source applications we have in the distribution.

The reason that glint does not see source rpms is due to that SRPMS are not stored in any of the RPM databases. This makes it almost impossible to tell if you have installed an src.rpm before or are over-writing an older version. Thus you will need to use the plain rpm command to install these items.

rpm -ivh < filename > will install the source code into the directory that the maintainer of that SRPM used. The data in src.rpms packaged by Red Hat are installed into
/usr/src/redhat by default.

Rebuilding and improving on rpms is beyond the scope of this answer. The book Maximum RPM and the man pages are good sources of information on this.

E.9.14 Linux recognizes my CDROM, but when I try to mount it, I get "mount failed" What do I do?

If your system was installed properly, simply typing mount /mnt/cdrom should work. If it does not, you must edit your /etc/fstab file. Here is an example of entry in /etc/fstab:

/dev/hdc        /mnt/cdrom     iso9660 noauto,ro 0 0
CD-ROM device)  (directory)    (filesystem type and options)

To find out what the CD-ROM device is, type dmesg | less and scan it for information regarding your CD-ROM. If you wish to mount the CD-ROM without adding this to your
/etc/fstab:

        mount -t iso9660 /dev/hdc /mnt/cdrom

E.9.15 I have Linux installed on an IDE drive, and for whatever reason I need to boot from floppy. How can I boot my system from the install floppy?

If you have installed Linux onto an IDE hard-drive, you can boot from the installation floppy using the following method.

Insert the installation floppy and restart the machine. At the boot: prompt type the following:

vmlinuz root=/dev/hdXY
  [Example: vmlinuz root=/dev/hdb5 ]

Where X = is the Linux drive letter and Y is the partition on the drive you installed the root (/) partition to.

E.9.16 I can't get my Plug and Play card to work.

The 2.0.xx kernels do not directly support the Plug and Play (PNP) protocol. You will need to either disable PNP on the card (via jumpers or card setup tools). You can also change your boot method to use Loadlin.exe from Windows (as windows would then have set up the PNP hardware).

Finally you can try using the isapnptools programs. First, type this:

pnpdump > /etc/isapnp.conf

This will create a configuration file that you will need to edit to choose the settings used for each card. Then type isapnp /etc/isapnp.conf to set up the devices.

See /urlhttp://www.roestock.demon.co.uk/isapnptools/ for more information.

E.9.17 When the system boots up, I see a message that says I have unknown PCI hardware. What does this mean?

The error "unknown PCI device" can occur for several reasons. The first and most harmless one is that PCI isn't responding to Linux's queries in a way it understands, but Linux is able to keep going. The more common occurrence is that the system hangs on querying PCI bus cards and cannot get any further.

Since this is a hardware problem in the kernel, there is not much that RedHat can do except point you to the maintainer of that section of the kernel. They may be able to let you know what is going on, and may want to look at what hardware you do have in your system so they can better handle it in the future. The maintainer can be reached at:

linux-pcisupport@cck.uni-kl.de

Please include the following information:

E.9.18 Currently when the machine boots, LILO defaults to running Linux. I would like it to boot my other operating system. How can I accomplish this?

To change the default OS that Linux boots into, you will need to edit the /etc/lilo.conf file and change the order of the OS's that LILO looks at. In the following example we change the order of booting so that DOS gets booted by default instead of Linux.

pico /etc/lilo.conf

# here is the old version

boot=/dev/hda
map=/boot/map
install=/boot/boot.b
prompt
timeout=50
image=/boot/vmlinuz-2.0.31
        label=linux
        root=/dev/hda2
        read-only
other = /dev/hda1
        label = dos
        table = /dev/hda

# change it to the following:
boot=/dev/hda
map=/boot/map
install=/boot/boot.b
prompt
timeout=50
other = /dev/hda1
        label = dos
        table = /dev/hda
image=/boot/vmlinuz-2.0.31
        label=linux
        root=/dev/hda2
        read-only

Save your changes to the file and leave the editor. Run the command

/sbin/lilo -v

and the updated lilo will be written to the boot device. On a reboot, the machine will boot into DOS as default now instead of Linux, with a 50 second delay to give you time to choose linux at the boot prompt if you wish to boot to Linux.

E.9.19 For whatever reason, I need to use the rescue mode to edit a file. The editors complain about unknown window type. How can I edit files after I have booted from the rescue disk?

After selecting rescue mode and answering a few questions, you will get a root prompt (#). You will need to mount the Linux partition like is done in this example. The partitions below are an example only. You should change them to be appropriate for your system (sda1,sdb1,hda5, etc)

mkdir /mnt
mount /dev/sdb1 /mnt
cd /mnt/etc
export TERMCAP=/mnt/etc/termcap
vi filename

or, if you wish to use a different editor, such as pico, you may want to chroot /mnt.


Next Up Previous Contents Index