Installation Problems

Installation Problems

Some possible problems could occur during the installation, one of which is that you have no free space. Many Linux distributions will allow you to install it on an existing DOS partition, so you need to be careful with this one. Fortunately, the system will tell what kind of file system it is.

If you are having trouble booting from floppies, there are a few reasons for this. The first, which I saw quite often while I was in tech support, happens on new machines when the CMOS setting for the floppy does not match what really is in the drive. Even today, CMOS are delivered in which the default A: floppy is 5.25″ and 360K. If you have a 3.5″ 1.44MB, it won’t work right. The kernel might load, because it is just reading one sector after the other. However, the first time you have to seek to a particular sector, you’re hosed.

Today, I no do not really see a need to install from floppies unless you have a very old machine. All CD drives that I have seen in recent years are are bootable and every major Linux distribution comes with bootable CDs. So all you need to do it put the CD in the drive, power on and go!

Another common problem is simply that it is defective media. If you can install directly from the CD-ROM, this problem is unlikely to happen. It might also be that the CD-ROM is just dirty. Wipe it off with a clean dry, non-static cloth. If you have to create a boot floppy set, it’s best to have new floppies. The cost of a new box is worth the piece of mind.

If the system hangs during boot process, pay close attention to any messages that come up, particularly the last few. These messages might indicate what the problems are. Although it didn’t cause a hang, I had an Ethernet card and multi-port board that were both software configurable. Both configured to IRQ 10. Because the multi-port board was initialized first, my Ethernet card was inaccessible. Also check the Hardware HOWTO to make sure that your hardware is supported.

I have heard of out-of-memory errors. I have never experienced them myself because I have always installed my system with 32MB. The reason for this error is that you have very little RAM and a RAM disk is created during the install for the root file system. If the amount of RAM you have minus the ramdisk doesn’t leave enough for Linux, you’re in trouble. If you don’t have enough RAM, get more. RAM is not prohibitively expensive, and you cant run a productive system without it anyway.

If you run into hardware problems, strip the system to just the devices you need for the install. Remove the parallel/serial ports, sound cards, network cards, mouse cards, and anything else you don’t need for the installation. If the problem still continues, maybe one of the other cards is loose (like the SCSI host adapter). Remove and reseat all the cards. Check for conflicts of base address, IRQ, DMA, and, obviously, check that your hardware is supported.

SCSI devices are good things to have, but they can also be a major headache. I must admit that at times I almost have a too-relaxed attitude when installing. (Hey, I know what I am doing, right?) I often take for granted that the hardware is configured in a particular way, especially when other hardware of the same type is configured that way. This leads to problems!

I have seen systems in which a SCSI device is detected at all IDs, though no devices are there and no manufacturers ID is recognized. This happens when there is a device and HA at the same SCSI ID. The host adapter must be at ID 7. Every other device must be at a different, unique ID.

Maybe Linux detects errors on the disk, but the disk is known to be error free, which is often caused by either bad cables or incorrect termination. If your SCSI devices report timeouts, there may be a conflict with base address, DMA, or IRQ of the host adapter. I have also experienced timeouts when more than two devices on the bus are terminated. Remember that only the two devices at the physical end of the SCSI bus should be (must be) terminated. the SCSI devices report timeouts, this may also be due to bad connections or termination problems. However, it might also be due to conflicts with the I/O address or IRQ or DMA channels.

If you have obtained a version from the Internet, you can get read errors or “file not found” messages. I specifically mentioned the Internet because I have never seen it with commercial versions. This might be an indication of bad media (so instead, use new floppies) or that something went wrong with the transfer.

If you used ftp to copy the files from DOS/Windows machine, the transfer mode is normally ASCII by default. These files must be transferred in binary mode or they could result in messages like “tar: read error” or “gzip: not in gzip format.”

After installing, you may have some problems booting. You may see something like “non-system disk in drive,” “disk not bootable,” “NO OS,” etc., probably because your master boot block is corrupt. This can occur when installing onto machines with multiple operating systems. Some OSs have to be installed on the active partition. I have experienced this once, when I installed a second OS on a system only to find that the second simply overwrote the first.

The fdisk program on all PC-based OSs can read the partition table. It may not see what kind of partition it is, but it will see that something is there. On the other hand, the Linux fdisk can read the partition table and recognize what kind of partitions they are.

Another cause of this kind of message is that LILO or GRUB might not have been installed correctly. Therefore, you should boot from the boot/root floppy that you made for the installation. Once you get to the boot:, you can specify the root file system to be on the hard disk. Once there, check the /etc/lilo.conf file and install it again.

One of my favorite problems I saw both with Linux and other UNIX dialects. When you boot, another OS starts instead of Linux, or Linux is installed first and after installing the other OS, Linux is no longer accessible. In the first case, glossary>LILO or GRUB may not have been installed. If it was installed, maybe it was not installed in the master boot record; but Linux is not active, so the other OS boots. Maybe you simply configured glossary>LILO or GRUB so that OS will boot by default. Ctrl+Shift+Tab and often just Tab gives you the LILO or GRUB menu. (See appropriate man-page for more details.)

Depending on the distribution, you may be able to choose what type of file system on which you want to install. If you will remember from the section on file systems, Linux supports several different types of file systems, most of which you can install on.

If you are fortunate to have a system that understands the RedHat Package Manager Format (rpm), adding additional software is fairly easy. See the rpm man-page for more details.