If your disk is an IDE or EIDE drive, you should read the file /usr/src/linux/drivers/block/README.ide (part of the Linux kernel source code). This README contains many helpful hints about IDE drives. Many modern IDE controllers do translation between `physical' cylinders/heads/sectors, and `logical' ones.
SCSI disks are accessed by linear block numbers. The BIOS invents some `logical' cylinder/head/sector fiction to support DOS.
An IBM PC-compatible BIOS will usually not be able to access partitions which extend beyond 1024 logical cylinders, and will make booting a Linux kernel from such partitions using LILO problematic at best.
You can still use such partitions for Linux or other operating systems that access the controller directly.
It's recommend that you create at least one Linux partition entirely under the 1024 logical cylinder limit, and boot from that. The other partitions will then be okay.
Also there seems to be a bit of trouble with the newer Ultra-DMA drives. I haven't gotten the straight scoop on them—but they are becoming a very common problem at the SVLUG installfests. When you can get 8 to 12 Gig drives for $200 to $300 it's no wonder.
[Jim Dennis]
In general, this is very hard to do on unices because of their multitasking nature. Undelete functionality for the ext2fs file system is being worked on, but don't hold your breath.
There are a number of packages available which instead provide new commands for deleting and copying which move deleted files into a `wastebasket' directory. The files can be recovered until cleaned out automatically by background processing.
The Midnight Commander file manager provides an undelete facility that uses Ext2 file system library functions and an undelete directory for each file system. Commercial distribution packages of MC may or may not have this feature enabled, so be sure to look in the source code distribution for instructions on how to enable the undelete feature.
Alternatively, you can search the raw disk device which holds the file system in question. This is hard work, and you will need to be logged in as root to do this. But it can be done. Run grep on the raw device; e.g.:
grep -b 'bookmarks' /dev/hda |
If the data has not been overwritten, you should be able to recover it with a text editor.
[Dave Cinege, Daniel Novotny]
You can back up a directory hierarchy or complete file system to any media using GNU tar or cpio, the standard *nix tools for this purpose. tar seems to be the more commonly used program currently, and includes command line options to make compressed, incremental, and multi-volume backups. Complete information is contained in the documentation, which is in GNU Texinfo format.
The free program, Amanda, receives a lot of mentions on Usenet. Its home page is http://www.amanda.org/.
Several commercial backup utilities also exist. They are often included in commercial distributions.
Use the FIPS.EXE program, included with most Linux distributions,under MS-DOS.
GNU parted, a partition editor, is stable enough for non-guru, mere-mortal use with relative confidence. Source code for the latest version is at: ftp://ftp.gnu.org/pub/gnu/parted/. There's also a boot disk image for resizing root partitions and for running parted on non-Linux machines. The disk image may be easier for beginners. Building from source could require some extra configuration.
Parted also has tutorial-style, plain-text documentation for Linux and FAT (MS-DOS) file systems.
Also, some commercial distributions come with their own partitioning software, like Partition Magic.
Yes. There is defrag, a Linux file system defragmenter for ext2, Minix, and old-style ext file systems. It is available at ftp://metalab.unc.edu/pub/Linux/system/filesystems/defrag-0.70.tar.gz.
Users of the ext2 file system can probably do without defrag, because ext2 contains extra code to keep fragmentation reduced even in very full file systems.
To format a 3.5-inch, high density floppy:
$ fdformat /dev/fd0H1440 $ mkfs -t ext2 -m 0 /dev/fd0H1440 1440 |
For a 5.25 inch floppy, use fd0h1200 and 1200 as appropriate. For the B: drive use fd1 instead of fd0.
The -m 0 option tells mkfs.ext2 not to reserve any space on the disk for the superuser—usually the last 10% is reserved for root.
The first command performs a low-level format. The second creates an empty file system. You can mount the floppy like a hard disk partition and simply cp and mv files, etc.
Device naming conventions generally are the same as for other unices. They can be found in Matt Welsh's Installation and Getting Started guide. Refer to (``Where Can I Get the HOWTO's and Other Documentation?'') A more detailed and technical description is Linux Allocated Devices by H. Peter Anvin, hpa@zytor.com, which is included in LaTeX and ASCII form in the kernel source distribution (probably in /usr/src/kernel/Documentation/), as devices.tex and devices.txt.
The most recent Linux kernels support software RAID, and they will work with RAID disk controllers.
An automounter for NFS partitions is part of most Linux distributions.
In addition, several virtual file system projects exist. One of them, the Linux Logical Volume Manager, is located at http://linux.msede.com/lvm/.
Yes. One file system, ppdd, is archived at http://pweb.de.uu.net/flexsys.mtk/.
You may have a corrupted file system, probably caused by not shutting Linux down properly before turning off the power or resetting. You need to use a recent shutdown program to do this—for example, the one included in the util-linux package, available on sunsite and tsx-11.
If you're lucky, the program fsck (or e2fsck or xfsck as appropriate if you don't have the automatic fsck front-end) will be able to repair your file system. If you're unlucky, the file system is trashed, and you'll have to re-initialize it with mkfs (or mke2fs, mkxfs, etc.), and restore from a backup.
NB: don't try to check a file system that's mounted read/write—this includes the root partition, if you don't see
VFS: mounted root ... read-only |
at boot time.
When you boot (or enable swapping manually) you should see
Adding Swap: NNNNk swap-space |
If you don't see any messages at all you are probably missing
swapon -av |
(the command to enable swapping) in your /etc/rc.local or /etc/rc.d/* (the system startup scripts), or have forgotten to make the right entry in /etc/fstab:
/dev/hda2 none swap sw |
for example.
If you see:
Unable to find swap-space signature |
you have forgotten to run mkswap. See the manual page for details; it works much like mkfs.
Running, free in addition to showing free memory, should display:
total used free Swap: 10188 2960 7228 |
If typing ``cat /proc/swaps'' reveals only file or partition names, but no swap space information, then the swap file or partition needs re-initialization.
Use fdisk (as root) to determine which partition on a hard drive has been designated as the swap partition. The partition still needs to be initialized with mkswap before enabling it with swapon.
[Andy Jefferson, Steve Withers]
In addition to a swap partition, Linux can also use a swap file. Some programs, like g++, can use huge amounts of virtual memory, requiring the temporary creation of extra space. To install an extra 64 MB of swap space, for example, use the following shell commands:
# dd if=/dev/zero of=/swap bs=1024 count=65535 # mkswap /swap # swapon /swap |
The count= argument to dd determines how big the swap file will be. In this example the swap file's name is /swap, but the name and location are, generally, arbitrary, depending only on the file system's available space and your having write permissions in the directory.
When you don't need the swap space any more, remove it with the following statements:
# swapoff /swap # rm /swap |
Take a look also at the Installation HOWTO and Installation & Getting Started for detailed instructions.
If that still doesn't provide enough swap space, refer to (``How Can I Have More Than 128Mb of Swap?'')
The lilo program (not the complete LILO package), uses the command line option -u to uninstall the LILO boot loader. You have to supply the device name of the device you installed LILO on, for example:
lilo -u /dev/hda |
This rewrites the original, pre-LILO master boot record back to the first hard drive, from the boot record saved in /boot/boot.0300. If you installed LILO to a partition as a secondary boot loader, for example, /dev/hda1, lilo re-installs the original boot sector from the save file /boot/boot.0301. Refer to the lilo manual page for details. Thanks to Villy Kruse for reminding me to update this answer.
If you have an earlier version of LILO, you will have to use the DOS (MS-DOS 5.0 or later, or OS/2) FDISK /MBR (which is not documented). This will restore a standard MS-DOS Master Boot Record. If you have DR-DOS 6.0, go into FDISK.EXE in the normal way and then select the Re-write Master Boot Record option.
If you don't have MS-DOS or DR-DOS, you need to have the boot sector that LILO saved when you first installed it. You did keep that file, didn't you? It's probably called boot.0301 or some such. Type:
dd if=boot.0301 of=/dev/hda bs=445 count=1 |
(or /dev/sda if you're using a SCSI disk). This may also wipe out your partition table, so beware! If you're desperate, you could use
dd if=/dev/zero of=/dev/hda bs=512 count=1 |
This will erase your partition table and boot sector completely: you can then reformat the disk using your favorite software. But this will render the contents of your disk inaccessible—you'll lose it all unless you're an expert.
Note that the DOS MBR boots whichever (single!) partition is flagged as ``active.'' You may need to use fdisk to set and clear the active flags on partitions appropriately.
The system call to format a floppy can only be done as root, regardless of the permissions of /dev/fd0*. If you want any user to be able to format a floppy, try getting the fdformat2 program. This works around the problems by being setuid to root.
Refer to (``EXT2-fs: warning: mounting unchecked file system.)
Remount it. If /etc/fstab is correct, you can simply type:
mount -n -o remount / |
If /etc/fstab is wrong, you must give the device name and possibly the type, too: e.g.
mount -n -o remount -t ext2 /dev/hda2 / |
To understand how you got into this state, see, (``EXT2-fs: warning: mounting unchecked file system.'')
None of the files in /proc are really there—they're all, ``pretend,'' files made up by the kernel, to give you information about the system and don't take up any hard disk space.
/proc/kcore is like an ``alias'' for the memory in your computer. Its size is the same as the amount of RAM you have, and if you read it as a file, the kernel does memory reads.
The option to allow disks with more than 1024 cylinders is only required as a workaround for a PC-compatible BIOS misfeature and should be turned `off' under Linux. For older Linux kernels you need to turn off most of the `advanced BIOS' options—all but the one about scanning the bus for bootable devices.
The journalling file system, named Reiserfs has just been released from testing. It is said to make Linux even faster than Linux with the Ext2 file system installed. Complete information is available at http://devlinux.org/namesys/.