Replacing all drives in a Terastation Live

This manual describes step by step the replacement of all drives in a Terastation Live. It can eventually applied also to other Terastations. The requirements are only a RAID-System based on mdadm and a XFS file system.

This document not only outlines the steps necessary to swap the drives but to also upgrade your Terastation's storage capacity without the need to backup all of its data. It is, however, important to note that you undertake the following with the potential risk of losing your data.

=Setup=

Buffalo Terastation comprising 4x Samsung SP2504C 250GB SATA-II

New drives: 4x Samsung HD501LJ 500GB SATA-II

NOTE: 4x Samsung HD103SJ 1000GB SATA-II drives have also been used

Necessary tools:


 * Java Runtime Environment
 * acp_commander
 * mdadm
 * xfs_growfs
 * fdisk

Necessary connectivity:


 * You will need console (telnet) access to your TeraStation in order to complete this process. If you are not familiar with how to gain this access, please read (at least the first part of) the "Open Stock Firmware" page here on NAS-Central.

We often start with a full array as shown by the 'df -h' command: root@DATEN:~# df -h Filesystem               Size      Used Available Use% Mounted on /dev/md1                481.6M    306.9M    174.7M  64% / /dev/ram1               15.0M    116.0k     14.9M   1% /mnt/ram /dev/md0               281.0M     15.0M    265.9M   5% /boot /dev/md2               695.6G    631.4G     64.2G  91% /mnt/array1

=Replace drives=

fdisk -l /dev/sda Disk /dev/sda: 250.0 GB, 250059350016 bytes 255 heads, 63 sectors/track, 30401 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes
 * Shut down the Terastation down via web interface, command line or power button
 * Start with the 4th drive, remove it and replace it with another disc of same size or bigger (you can follow the instructions from the Buffalo Terastation manual.
 * Start the Terastation and it will show you a failed drive - don't panic :-)
 * Now you have to gain root console access via telnet or ssh. It is important that you do not try to repair the raid(s) from the web interface! 1
 * Take a look at the partition table of a working hard drive like /dev/sda

Device Boot     Start         End      Blocks   Id  System /dev/sda1              1          37      297171   83  Linux /dev/sda2             38          99      498015   83  Linux /dev/sda4            100       30401   243400815    5  Extended /dev/sda5            100         116      136521   82  Linux swap /dev/sda6            117       30390   243175873+  83  Linux fdisk /dev/sdd
 * Create an identical partition table using fdisk on the newly inserted drive (/dev/sdd). This can be done through the command:
 * Important - If you are upgrading to larger drives, you will want to extend the size of the extended partition and partition 6 to the maximum available size (last cylinder number).
 * Create /dev/sdd1 & sdd2 as primary partitions number 1 & 2, respectively. Use starting & ending cylinder numbers to match /dev/sda1 & sda2 from your 'fdisk -l /dev/sda' output.
 * Create /dev/sdd4 as an extended partition using the remaining cylinders.
 * Create /dev/sdd5 & sdd6 as logical partitions. Use the same starting & ending cylinder number matching--as shown in your 'fdisk -l /dev/sda' output--for these partitions.
 * Change partition number 5 to a Linux Swap partition. Press 'l' (lowercase L) within fdisk to verify that Linux Swap partitions are assigned hexadecimal 82. And then press 't' to change partition 5 to the appropriate type/number.

Disk /dev/sdd: 500.1 GB, 500107862016 bytes 255 heads, 63 sectors/track, 60801 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot     Start         End      Blocks   Id  System /dev/sdd1              1          37      297171   83  Linux /dev/sdd2             38          99      498015   83  Linux /dev/sdd4            100       60801   487588815    5  Extended /dev/sdd5            100         116      136521   82  Linux swap /dev/sdd6            117       60801   487452231   83  Linux mkswap /dev/sdd5 mdadm -a /dev/md0 /dev/sdd1 mdadm -a /dev/md1 /dev/sdd2 mdadm -a /dev/md2 /dev/sdd6 root@daten# mdadm --detail /dev/md0
 * apply mkswap to the fifth partition
 * now rebuild all three raids (0-2), the steps are as follows 1
 * now you can check if this was done correctly with

/dev/md0: Version : 00.90.03 Creation Time : Wed Jun 13 15:47:40 2007 Raid Level : raid1 Array Size : 297088 (290.17 MiB 304.22 MB) Device Size : 297088 (290.17 MiB 304.22 MB) Raid Devices : 4 Total Devices : 4 Preferred Minor : 0 Persistence : Superblock is persistent

Update Time : Tue Oct 30 07:39:29 2007 State : clean Active Devices : 4 Working Devices : 4 Failed Devices : 0 Spare Devices : 0

UUID : ec67ebdb:01b029ce:000bb777:9d5fd144 Events : 0.730

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 3      8       49        3      active sync   /dev/sdd1
 * The rebuilding of arrays 0 & 1 runs quickly, but the large array can take several hours. If you access your NAS' web interface, the home page will contain a link indicating that it is rebuilding array 1. By following that link, you can get an approximation of the rebuild time.
 * Important - do not replace any other drives until the array rebuilding is done.
 * Repeat all these steps, including the shutdown and restart for all drives.

=Growing the array and file system=

mdadm --grow /dev/md2 -z max root@DATEN:~# mdadm --detail /dev/md2 /dev/md2: Version : 00.90.03 Creation Time : Wed Jun 13 08:29:22 2007 Raid Level : raid5 Array Size : 1462356480 (1394.61 GiB 1497.45 GB) Device Size : 487452160 (464.87 GiB 499.15 GB) Raid Devices : 4 Total Devices : 4 Preferred Minor : 2 Persistence : Superblock is persistent
 * 1 After all drives are replaced and the rebuild finished you can grow your storage array using the following:
 * the array daemon starts a resync subsequently after growing the array

Update Time : Wed Oct 31 08:36:13 2007 State : active, resyncing Active Devices : 4 Working Devices : 4 Failed Devices : 0 Spare Devices : 0

Layout : left-symmetric Chunk Size : 64K

Rebuild Status : 50% complete

UUID : 05c76307:69dff621:a11d513a:e4478523 Events : 0.1320002

Number  Major   Minor   RaidDevice State 0      8        6        0      active sync   /dev/sda6 1      8       22        1      active sync   /dev/sdb6 2      8       38        2      active sync   /dev/sdc6 3      8       54        3      active sync   /dev/sdd6 xfs_growfs /mnt/array1/ root@DATEN:~# df -h Filesystem               Size      Used Available Use% Mounted on /dev/md1                481.6M    307.0M    174.6M  64% / /dev/ram1               15.0M    116.0k     14.9M   1% /mnt/ram /dev/md0               281.0M     15.0M    265.9M   5% /boot /dev/md2                 1.4T    631.4G    763.1G  45% /mnt/array1
 * this step will also take several hours (about 5 hours in this case)
 * the last step is growing the file system which runs very quick, just run
 * and you can do that before the resync finished.
 * so we end up with a enlarged array without any data loss or need to backup the data! :-)

root@DATEN:~# cat /proc/mdstat Personalities : [linear] [raid0] [raid1] [raid5] [raid4] md2 : active raid5 sda6[0] sdb6[1] sdd6[3] sdc6[2] 1462356480 blocks level 5, 64k chunk, algorithm 2 [4/4] [UUUU] [============>........] resync = 64.8% (315880576/487452160) finish=191.9min speed=14865K/sec
 * you can watch the progress by using the web interface or looking into /proc/mdstat

md1 : active raid1 sda2[0] sdb2[1] sdd2[3] sdc2[2] 497920 blocks [4/4] [UUUU]

md0 : active raid1 sda1[0] sdb1[1] sdd1[3] sdc1[2] 297088 blocks [4/4] [UUUU]

unused devices:

=Notes=
 * [1] If you have an empty Terastation, or you have a good backup of the data you keep on there, you can skip the parts of these instructions that reference /dev/md2, including growing the array and filesystem. If you do skip those parts, you can then create your new storage array in the web interface, but as mentioned before, DO NOT try to fix any old arrays from the web interface.

=Links=


 * Mikas Blog
 * fdisk-Howto