mdadm

from Wikipedia, the free encyclopedia
mdadm
Basic data

Maintainer Jes Sorensen, formerly Neil Brown
Publishing year 2001
Current  version 4.0
(January 9, 2017)
operating system Linux
programming language C.
category Disk management
License GPL ( Free Software )
German speaking No
neil.brown.name/blog/mdadm

mdadm (for m ultiple d isk adm inistration, formerly also mdctl ) is a utility program for Linux for managing a software RAID . The program can be used to create, configure, monitor and delete RAID networks. mdadm is released as free software under the GNU General Public License (GPL).

Features

mdadm creates so-called multiple devices (MD for short) from various block devices (such as an entire hard drive, a single partition or a USB stick). The MD can then be used as a block device by the MD device driver . mdadm supports the following types of MDs:

  • Linear: concatenation of several partitions
  • Multipath: No RAID, but a mapping of a file to two different paths on the same partition ( mirroring )
  • Faulty: Emulates a faulty RAID system for test cases
  • Level 0 (Block Level Striping): Concatenation of several small block devices into one large one.
  • Level 1 (Mirror): Mirroring a disc
  • Level 4: Like level 0, but with an additional device for parity bits (increased reliability ).
  • Level 5: Like level 4, but the parity bits are distributed over all devices.
  • Level 6: Like level 5, but with two additional devices for two independent parity bits per segment (further increased reliability ).
  • Level 10: Combination of level 0 over level 1.

More about the various RAID levels can also be found in the article RAID # The common RAID levels in detail .

Before the Linux - Kernel 2.6 of the kernel could mdadm the MDs as / dev / MDN (where n is a number between 0 and 99 is) include 2.4 to as / dev / md / name, said name a freely selectable name. These devices were not partitionable . The kernel 2.6. introduced new partitionable arrays . The devices are now called / dev / md_d2p3 (the 3rd partition of the 2nd MD). As of Kernel 2.6.28, the old MDs can also be partitioned. They are then called / dev / md2p3 (the 3rd partition of the 2nd MD).

Booting from a software raid

Basically, the problem with booting a kernel that is on an MD is that the BIOS or bootloader does not support the RAID algorithms used. In the meantime, with GRUB2 and its modules, the possibility has been created to provide simple software RAID support in the bootloader. It is now also possible to boot a kernel from an LVM volume from a soft raid. The boot loader and its modules are located in the first sectors of hard disks partitioned for MBR and in a separate partition for hard disks partitioned for GPT.

Mdmpd

Mdmpd is a computer program for GNU / Linux - operating systems . It is part of the mdadm package and was originally developed by Red Hat .

The program is loaded as a daemon when the system is started and then used to monitor RAID networks. The name of the program mdmpd is a synonym for engl. daemon to monitor MD m ulti p ath d evices (German background service for monitoring multiple transmissions on hard disks ).

Storage solutions in companies often require several options for communication with a single drive within a RAID array. In this way, in the event of a fault in the controller to which a drive is connected, the system can easily communicate with the drive via another controller and thus remains fully functional. This solution is also called multipath disk access (access to drives over several channels) and is used, for example, with SAS hard drives. This type of access is implemented using the software raid functionality of the Linux kernel, also known as the md driver (multiple disk).

functionality

The component of the md driver in the kernel only handles I / O requests to the corresponding drive and error handling of the active channel. There is no check to determine whether a previously defective channel is functional again.

When the system is started, the daemon carries out an initial check of the RAID array and saves this status, after which the daemon runs in the background and waits for further notifications from the kernel.

After receiving such a message, the daemon checks whether the active channel to a drive has failed. A request is then sent through this channel every 15 seconds until the failure has been rectified by changing the channel. As soon as the problem has been resolved, the daemon adds the (now no longer active) channel to the original RAID as a backup channel.

When using procfs , a dynamic file system for storing system-relevant data, all configured RAID networks and other information about their status are saved in the virtual file / proc / mdstat. mdmpd needs this information to localize the RAID system, to identify the components of the RAID system and to notify you about relevant events.

Web links

Individual evidence

  1. ↑ a b ANNOUNCE: mdadm 4.0 - A tool for managing md Soft RAID under Linux
  2. ↑ mdadm on Neil Brown's homepage ( memento of the original from May 3, 2013 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. @1@ 2Template: Webachiv / IABot / neil.brown.name
  3. ↑ Man page
  4. ↑ http://www.gnu.org/software/grub/manual/grub.html#Changes-from-GRUB-Legacy