--- name: "ext2prepare" section: "8" platform: "linux" package: "ext2resize" version: "1.1.19-9" release: "Debian 4.0 (Etch)" platforms: ["linux"] url: "https://new.carta.tech/ext2prepare" --- # ext2prepare(8) ext2 filesystem resizing preparation tool ## NAME ext2prepare - ext2 filesystem resizing preparation tool ## SYNOPSIS **ext2prepare** \[**-d**\] \[**-f**\] \[**-q**\] \[**-v**\] \[**-V**\] *device size*\[**b**\|**K**\|**M**\|**G**\|**T**\]\] ## DESCRIPTION The **ext2prepare** command modifies an unmounted ext2 filesystem on *device* so that [**ext2online**(8)](/ext2online/8.md) program can later resize the mounted filesystem to be at least as large as *size* although the actual maximum size will be some larger multiple of various filesystem parameters. While **ext2prepare** must be run on an unmounted filesystem, *device* does not have to currently be as large as *size*. This allows one to later increase the size of *device* and do the resize while the filesystem is mounted. If the *size* parameter does not have a modifier, it will be taken to be in ext2 filesystem blocks (which can be 1k, 2k, or 4k - use [**dumpe2fs**(8)](/dumpe2fs/8.md) to find out more information about the current filesystem). The modifiers **b**, **K**, **M**, **G**, or **T** mean the *size* parameter is given in 512-byte blocks, kilo-, mega-, giga-, or terabytes respectively. The **ext2prepare** program does not change the size of the actual device. If you wish to enlarge a filesystem, you must make sure you expand the underlying device first. This can be done online for Logical Volumes by using [**lvextend**(8)](/lvextend/8.md) from the LVM package, or possibly via **md** or **RAID** facilities in the hardware if you are not using partitions. It is **not** possible to do this by using the [**fdisk**(8)](/fdisk/8.md) family of tools to extend a partition while it is mounted (at least the author has not been able to successfully do this). Because of the original design of the ext2 filesystem did not have online resizing in mind, there are certain limitations to the amount of resizing that can be done while the filesystem is mounted, if you haven't done any preparation for the resize. The default block size for ext2 was 1k blocks until v1.15 of e2fsprogs (1999), where it changed to 4k blocks for filesystems larger than 512MB. With no filesystem preparation, it is always possible to resize to the next 256MB boundary for 1k filesystems, the next 2GB boundary for for 2k filesystems, and the next 16GB boundary for 4k filesystems. By using **ext2prepare** on an unmounted filesystem, it is possible for [**ext2online**(8)](/ext2online/8.md) to increase the size of a mounted ext2 filesystem to at least the given *size* by reserving blocks for group descriptors within the ext2 filesystem structure. **ext2prepare** does this by moving metadata blocks while the filesystem is unmounted, and then allocating the freed blocks to a reserved inode so that [**ext2online**(8)](/ext2online/8.md) can later use these blocks to safely enlarge the filesystem while the filesystem is in use. ## OPTIONS - **-d**, **--debug** Turn on debugging messages. - **-f**, **--force** Force preparation without first checking the filesystem state. - **-q**, **--quiet** Do not print anything but error messages. - **-v**, **--verbose** Turn on normal verbose status messages. - **-V**, **--version** Print the version number and exit. ## RETURN VALUES 0 Resizing successful - 1 Error in options or command-line parameters. - 2 Error while preparing filesystem. ## EXAMPLES The following example shows how to test **ext2prepare** with a spare partition. First a filesystem of 32MB is created on the device. The filesystem is then prepared to grow to a maximum size of 10GB, mounted, and the size is verified. The filesystem is then extended to fill the device (the default action when no size is given), and the new size is verified. mke2fs /dev/vg0/lvtest 32768\ ext2prepare /dev/vg0/lvtest 10G\ mkdir /mnt/test\ mount -o debug,check=strict /dev/vg0/lvtest /mnt/test\ df /mnt/test\ ext2online -d -v /dev/vg0/lvtest\ df /mnt/test By using the *debug* and *check=strict* options to mount, as well as the **-d** and **-v** options to **ext2prepare**, we will see the user-space status messages, and the kernel messages will be logged by syslog. ## WARNING Note that resizing a mounted filesystem is inherently dangerous and may corrupt filesystems, although no errors resulting in data loss have ever been reported to the author. In theory online resizing should work fine with arbitrarily large filesystems, but it has not yet been tested by the author on a filesystem larger than 11GB. **Use with caution.** Backups are **always** a good idea, because your disk may fail at any time, you delete files by accident, or your computer is struck by a meteor. ## AUTHOR The **ext2prepare** program was written by Andreas Dilger \ using the ext2 resizing tools developed by Lennert Buytenhek \. The ext2online kernel patches are available as part of the GNU ext2resize package at \. ## BUGS Because **ext2prepare** uses a reserved inode to store blocks needed for a future resize (so that they can't be accidentally used by another program), [**e2fsck**(8)](/e2fsck/8.md) needs to be slightly modified to understand this new reserved inode. If an unmodified [**e2fsck**(8)](/e2fsck/8.md) is run on a filesystem that has been changed with **ext2prepare**, it will complain about the reserved inode and free all of the blocks associated with this inode. If this happens, it will not be possible to resize the filesystem past the default limits given previously. However, the filesystem itself is not affected in any way by these events. The ext2resize programs do **not** work on big-endian machines (Alpha, SPARC, PPC, etc). ## COPYRIGHT **ext2prepare** is (C) Copyright 1999, 2000, 2001 by Andreas Dilger and may be distributed under the terms of the GNU General Public License. ## SEE ALSO [**dumpe2fs**(8)](/dumpe2fs/8.md) [**ext2online**(8)](/ext2online/8.md) [**ext2resize**(8)](/ext2resize/8.md) [**e2fsck**(8)](/e2fsck/8.md) [**lvextend**(8)](/lvextend/8.md)