6.3. Managing Software RAID

This section discusses software RAID configuration and management after the installation, and covers the following topics:
  • Reviewing existing software RAID configuration.
  • Creating a new RAID device.
  • Replacing a faulty device in an array.
  • Adding a new device to an existing array.
  • Deactivating and removing an existing RAID device.
  • Saving the configuration.
All examples in this section use the software RAID configuration from the previous section.

6.3.1. Reviewing RAID Configuration

When a software RAID is in use, basic information about all presently active RAID devices are stored in the /proc/mdstat special file. To list these devices, display the content of this file by typing the following at a shell prompt: 
cat /proc/mdstat
To determine whether a certain device is a RAID device or a component device, run the command in the following form as root: 
mdadm --query device…
In order to examine a RAID device in more detail, use the following command: 
mdadm --detail raid_device…
Similarly, to examine a component device, type: 
mdadm --examine component_device…
While the mdadm --detail command displays information about a RAID device, mdadm --examine only relays information about a RAID device as it relates to a given component device. This distinction is particularly important when working with a RAID device that itself is a component of another RAID device.
The mdadm --query command, as well as both mdadm --detail and mdadm --examine commands allow you to specify multiple devices at once.

Example 6.1. Reviewing RAID configuration

Assume the system uses configuration from Figure 6.7, “Sample RAID Configuration”. You can verify that /dev/md0 is a RAID device by typing the following at a shell prompt: 
~]# mdadm --query /dev/md0
/dev/md0: 125.38MiB raid1 2 devices, 0 spares. Use mdadm --detail for more detail.
/dev/md0: No md super block found, not an md component.
As you can see, the above command produces only a brief overview of the RAID device and its configuration. To display more detailed information, use the following command instead: 
~]# mdadm --detail /dev/md0
/dev/md0:
        Version : 0.90
  Creation Time : Tue Jun 28 16:05:49 2011
     Raid Level : raid1
     Array Size : 128384 (125.40 MiB 131.47 MB)
  Used Dev Size : 128384 (125.40 MiB 131.47 MB)
   Raid Devices : 2
  Total Devices : 2
Preferred Minor : 0
    Persistence : Superblock is persistent

    Update Time : Thu Jun 30 17:06:34 2011
          State : clean
 Active Devices : 2
Working Devices : 2
 Failed Devices : 0
  Spare Devices : 0

           UUID : 49c5ac74:c2b79501:5c28cb9c:16a6dd9f
         Events : 0.6

    Number   Major   Minor   RaidDevice State
       0       3        1        0      active sync   /dev/hda1
       1       3       65        1      active sync   /dev/hdb1
Finally, to list all presently active RAID devices, type: 
~]$ cat /proc/mdstat
Personalities : [raid0] [raid1]
md0 : active raid1 hdb1[1] hda1[0]
      128384 blocks [2/2] [UU]
      
md1 : active raid0 hdb2[1] hda2[0]
      1573888 blocks 256k chunks

md2 : active raid0 hdb3[1] hda3[0]
      19132928 blocks 256k chunks

unused devices: <none>

6.3.2. Creating a New RAID Device

To create a new RAID device, use the command in the following form as root: 
mdadm --create raid_device --level=level --raid-devices=number component_device…
This is the simplest way to create a RAID array. There are many more options that allow you to specify the number of spare devices, the block size of a stripe array, if the array has a write-intent bitmap, and much more. All these options can have a significant impact on the performance, but are beyond the scope of this document. For more detailed information, refer to the CREATE MODE section of the mdadm(8) manual page.

Example 6.2. Creating a new RAID device

Assume that the system has two unused SCSI disk drives available, and that each of these devices has exactly one partition of the same size: 
~]# ls /dev/sd*
/dev/sda  /dev/sda1  /dev/sdb  /dev/sdb1
To create /dev/md3 as a new RAID level 1 array from /dev/sda1 and /dev/sdb1, run the following command: 
~]# mdadm --create /dev/md3 --level=1 --raid-devices=2 /dev/sda1 /dev/sdb1
mdadm: array /dev/md3 started.

6.3.3. Replacing a Faulty Device

To replace a particular device in a software RAID, first make sure it is marked as faulty by running the following command as root: 
mdadm raid_device --fail component_device
Then remove the faulty device from the array by using the command in the following form: 
mdadm raid_device --remove component_device
Once the device is operational again, you can re-add it to the array: 
mdadm raid_device --add component_device

Example 6.3. Replacing a faulty device

Assume the system has an active RAID device, /dev/md3, with the following layout (that is, the RAID device created in Example 6.2, “Creating a new RAID device”): 
~]# mdadm --detail /dev/md3 | tail -n 3
    Number   Major   Minor   RaidDevice State
       0       8        1        0      active sync   /dev/sda1
       1       8       17        1      active sync   /dev/sdb1
Imagine the first disk drive fails and needs to be replaced. To do so, first mark the /dev/sdb1 device as faulty: 
~]# mdadm /dev/md3 --fail /dev/sdb1
mdadm: set /dev/sdb1 faulty in /dev/md3
Then remove it from the RAID device: 
~]# mdadm /dev/md3 --remove /dev/sdb1
mdadm: hot removed /dev/sdb1
As soon as the hardware is replaced, you can add the device back to the array by using the following command: 
~]# mdadm /dev/md3 --add /dev/sdb1
mdadm: added /dev/sdb1

6.3.4. Extending a RAID Device

To add a new device to an existing array, use the command in the following form as root: 
mdadm raid_device --add component_device
This will add the device as a spare device. To grow the array to use this device actively, type the following at a shell prompt: 
mdadm --grow raid_device --raid-devices=number

Example 6.4. Extending a RAID device

Assume the system has an active RAID device, /dev/md3, with the following layout (that is, the RAID device created in Example 6.2, “Creating a new RAID device”): 
~]# mdadm --detail /dev/md3 | tail -n 3
    Number   Major   Minor   RaidDevice State
       0       8        1        0      active sync   /dev/sda1
       1       8       17        1      active sync   /dev/sdb1
Also assume that a new SCSI disk drive, /dev/sdc, has been added and has exactly one partition. To add it to the /dev/md3 array, type the following at a shell prompt: 
~]# mdadm /dev/md3 --add /dev/sdc1
mdadm: added /dev/sdc1
This will add /dev/sdc1 as a spare device. To change the size of the array to actually use it, type: 
~]# mdadm --grow /dev/md3 --raid-devices=3

6.3.5. Removing a RAID Device

To remove an existing RAID device, first deactivate it by running the following command as root: 
mdadm --stop raid_device
Once deactivated, remove the RAID device itself: 
mdadm --remove raid_device
Finally, zero superblocks on all devices that were associated with the particular array: 
mdadm --zero-superblock component_device…

Example 6.5. Removing a RAID device

Assume the system has an active RAID device, /dev/md3, with the following layout (that is, the RAID device created in Example 6.4, “Extending a RAID device”): 
~]# mdadm --detail /dev/md3 | tail -n 4
    Number   Major   Minor   RaidDevice State
       0       8        1        0      active sync   /dev/sda1
       1       8       17        1      active sync   /dev/sdb1
       2       8       33        2      active sync   /dev/sdc1
In order to remove this device, first stop it by typing the following at a shell prompt: 
~]# mdadm --stop /dev/md3
mdadm: stopped /dev/md3
Once stopped, you can remove the /dev/md3 device by running the following command: 
~]# mdadm --remove /dev/md3
Finally, to remove the superblocks from all associated devices, type: 
~]# mdadm --zero-superblock /dev/sda1 /dev/sdb1 /dev/sdc1

6.3.6. Preserving the Configuration

By default, changes made by the mdadm command only apply to the current session, and will not survive a system restart. At boot time, the mdmonitor service reads the content of the /etc/mdadm.conf configuration file to see which RAID devices to start. If the software RAID was configured during the graphical installation process, this file contains directives listed in Table 6.1, “Common mdadm.conf directives” by default.

Table 6.1. Common mdadm.conf directives

Option Description
ARRAY
Allows you to identify a particular array.
DEVICE
Allows you to specify a list of devices to scan for a RAID component (for example, “/dev/hda1”). You can also use the keyword partitions to use all partitions listed in /proc/partitions, or containers to specify an array container.
MAILADDR Allows you to specify an email address to use in case of an alert.
To list what ARRAY lines are presently in use regardless of the configuration, run the following command as root: 
mdadm --detail --scan
Use the output of this command to determine which lines to add to the /etc/mdadm.conf file. You can also display the ARRAY line for a particular device: 
mdadm --detail --brief raid_device
By redirecting the output of this command, you can add such a line to the configuration file with a single command: 
mdadm --detail --brief raid_device >> /etc/mdadm.conf

Example 6.6. Preserving the configuration

By default, the /etc/mdadm.conf contains the software RAID configuration created during the system installation: 
# mdadm.conf written out by anaconda
DEVICE partitions
MAILADDR root
ARRAY /dev/md0 level=raid1 num-devices=2 UUID=49c5ac74:c2b79501:5c28cb9c:16a6dd9f
ARRAY /dev/md1 level=raid0 num-devices=2 UUID=76914c11:5bfa2c00:dc6097d1:a1f4506d
ARRAY /dev/md2 level=raid0 num-devices=2 UUID=2b5d38d0:aea898bf:92be20e2:f9d893c5
Assuming you have created the /dev/md3 device as shown in Example 6.2, “Creating a new RAID device”, you can make it persistent by running the following command: 
~]# mdadm --detail --brief /dev/md3 >> /etc/mdadm.conf