A model E or F appliance is shipped with two Solid State Disks (SSDs). When a disk becomes faulty, you can replace it using the mwg-raid tool.
The disks that the appliance is shipped with are named sda and sdb. When this configuration is modified later on, for example, when an HDD or USB drive is added, other disk names can also be used, for example, sdc or sdd.
The disks are partitioned and configured for RAID-1.
- Log on to the appliance where you want to replace a disk from a local system console or remotely using SSH.
When connecting with SSH, consider using a terminal multiplexer to ensure that the replacement procedure does not fail due to an unstable or broken SSH connection.
You can use the tmux multiplexer that the appliance has installed.
Run the following command to find out about the names of the disks that are currently installed:
fdisk -l | grep sd
The command outputs lines showing, for example, sda, sdb, sdc, and sdd as disk names.
Disk /dev/sda: 32.0 GB, 32017047552 bytes, 62533296 sectors
. . .
Disk /dev/sdb: 480.1 GB, 480103981056 bytes, 937703088 sectors
. . .
Disk /dev/sdc: 480.1 GB, 480103981056 bytes, 937703088 sectors
. . .
Disk /dev/sdd: 31.4 GB, 31406948352 bytes, 61341696 sectors
. . .
Here is the complete output.
[root@WBG-4500E ~]# fdisk -l | grep sd Disk /dev/sda: 32.0 GB, 32017047552 bytes, 62533296 sectors /dev/sda1 * 0 1642495 821248 17 Hidden HPFS/NTFS Disk /dev/sdb: 480.1 GB, 480103981056 bytes, 937703088 sectors /dev/sdb1 2048 21483519 10740736 fd Linux raid autodetect /dev/sdb2 21483520 937701375 458108928 fd Linux raid autodetect Disk /dev/sdc: 480.1 GB, 480103981056 bytes, 937703088 sectors /dev/sdc1 2048 21483519 10740736 fd Linux raid autodetect /dev/sdc2 21483520 937701375 458108928 fd Linux raid autodetect Disk /dev/sdd: 31.4 GB, 31406948352 bytes, 61341696 sectors /dev/sdd1 * 63 2242783 1121360+ c W95 FAT32 (LBA) [root@WBG-4500E ~]#
- Find out which disk is located where with the following command:
mwg-raid locate sd<x>
Run this command with different letters for <x>, for example, a, b, c, or d, depending on what you have found out about the names of the disks that are installed.
The LED on the disk with the specified letter, for example, the LED on the disk named sda, sdb, sdc, or sdd, starts blinking.
If none of the letters in the names returned by the fdisk -l command causes the LED on the faulty disk to start blinking, use a workaround to complete the procedure, see Use a Workaround to Replace a Disk on a Model E or F Appliance. Otherwise continue with step 4.
Remove the disk from the RAID configuration before you remove it physically with this command:
In the command, specify the appropriate letter for <x>, for example, a, b, c, or d, depending on which disk you want to remove.
This removes the specified disk from all RAID volumes where it was a member. The LED on the disk turns to continuous red, as the disk status must be faulty before a disk can be physically removed.
Remove the old disk from the slot on the appliance hardware platform and insert the new.
Add the new disk to the RAID configuration with this command:
mwg-raid add sd<x>
For <x> specify the letter that was used for the old disk, for example, a, b, c, or d.
The new disk is added to the RAID configuration. The content of the disk that was removed is mirrored on the new disk. While this process is going on, the red LED blinks slowly.
- To check whether the replacement has been performed correctly, run this command:
The command outputs, for example, the following:
md0 raid1 clean md0:0 sda1 md0:1 sdb1 md1 raid1 clean md1:0 sda2 md1:1 sdb2
This shows that RAID-1 status has been set up again for the disks on the appliance.