• Home
• Page Info
• Changes
• Attachments
• More Actions: Raw Text Print View Delete Cache ------------------------ Check Spelling Like Pages Local Site Map ------------------------ Rename Page Delete Page ------------------------ Subscribe User ------------------------ Remove Spam Revert to this revision Package Pages Sync Pages ------------------------ Load Save SlideShow

Server RAID1 Install

Server basic RAID1 install

Case ID: SR1I-001

• NB: A more exhaustive set of test instructions to be used in KVM environments can be found on the BootDegradedRaid wiki page

1. Boot CD and run the CD self-check (then reboot)

2. Select Install Ubuntu Server or ""Install to Hard Disk""

3. Choose your language, country and keyboard layout
4. Set hostname: default - ubuntu
5. Confirm time zone detection.
6. Partitioning method: "Manual".
   1. RAID1 array for /
   2. RAID1 array for swap
   3. RAID1 array for /home (testing non-rootfs raid)
7. Select "Yes" to the "boot degraded?" question
8. User account: enter username and password
9. No software selection.
10. Reboot and login.
11. Make sure that the root and home file systems are mounted from md devices:

    • sudo mount

12. Make sure that the swap partition is mounted from a md device:

    • sudo cat /proc/swaps

13. Make sure that the raid arrays are working:

    • cat /proc/mdstat

14. Make sure that grub has been installed on both disks:

    • sudo apt-get install -y binutils
      for i in $(sudo mdadm -Q --detail $(df -P /boot | grep ^/dev/ | cut -d" " -f1) | grep " /dev/" | awk '{print $NF}' | sed -e 's/[0-9]$//'); do sudo dd if=$i bs=512 count=1 2>/dev/null | strings -a | grep -q GRUB && echo $i: ok || echo $i: FAIL; done

15. Make sure that the BOOT_DEGRADED setting is "TRUE" in /etc/initramfs-tools/conf.d/mdadm:

    • cat /etc/initramfs-tools/conf.d/mdadm

16. Test booting from a cold-degraded array:
    1. Poweroff the system
    2. Disconnect one of the disk (disk 2) - disk 1 connected, disk2 disconnected.
    3. Poweron the system
    4. Check that system boots correctly from the degraded RAID1 array on a single disk (note that you may have to wait up to 5 minutes for mdadm to time out and boot into degraded mode):

       • cat /proc/mdstat

    5. Poweroff the system

    6. If you simply disconnect the first disk (disk1) and reconnect the second disk (disk2) - disk 1 disconnected, disk 2 connected - you risk disk corruption; see bug 557429)

    7. Reconnect the second disk (disk2) - both disks now connected.
    8. Poweron the system
    9. Check that the system boots correctly (there should be no error or delay)

    10. Check the status of the raid arrays:

        cat /proc/mdstat

    11. All arrays should have been assembled completely again, possibly still syncing.

    12. You may have to add any missing devices back to the RAIDs manually. This is not a bug (see bug 791454)! A manual addition would be:

        sudo mdadm --add /dev/mdX /dev/MISSING-DEVICE

        Note that this may fail with a message requiring you to zero the superblock first, this is a result of an added check in mdadm 3.2, and should only happen on precise or later (see bug 943397).

    13. make sure that all disk arrays are synchronized before proceeding, if the array is doing a full re-sync, it may take a few minutes, use

        cat /proc/mdstat

    14. Poweroff the system.
    15. Disconnect the first disk (disk1) - disk 1 disconnected, disk 2 connected.
    16. Poweron the system.
    17. Check that system boots correctly from the degraded RAID1 array on a single disk:

        • cat /proc/mdstat

17. Test automatic syncing of degraded array:
    1. Poweroff the system
    2. Reconnect the first disk (so both are now connected)
    3. Poweron the system
18. Wait for both drives to be back in sync:

    • cat /proc/mdstat

19. Change "Do you want to boot degraded?" answer to "No":

    • sudo dpkg-reconfigure mdadm

20. Make sure that the BOOT_DEGRADED setting is "FALSE" in /etc/initramfs-tools/conf.d/mdadm:

    • cat /etc/initramfs-tools/conf.d/mdadm

21. Test booting from a cold-degraded array:
    1. Poweroff the system
    2. Disconnect one of the disk (disk 2) - disk 1 connected, disk2 disconnected.
    3. Poweron the system
    4. Check that on boot a question to enable and boot from a degraded array is asked.
    5. Say yes
    6. Check that system boots correctly from the degraded RAID1 array:

       • cat /proc/mdstat

    7. Poweroff the system
    8. Disconnect the first disk (disk1) and reconnect the second disk (disk2) - disk 1 disconnected, disk 2 connected.
    9. Poweron the system.
    10. Check that on boot a question to enable and boot from a degraded array is asked.
    11. Say yes
    12. Check that system boots correctly from the degraded RAID1 array:

        • cat /proc/mdstat

22. Re-add/sync the arrays again
    1. Poweroff the system
    2. Reconnect the first disk (so both are now connected)
    3. Poweron the system
    4. Add the missing drives back to the RAIDs:

       • sudo mdadm -a /dev/mdX /dev/MISSING-DEVICE

23. Test booting from a hot-degraded array:
    1. Remove (unplug/fail) one disk from the running system.
    2. Check if users/admin get a notification message and beep about the failing raid.
    3. Reboot, verify that system comes up degraded without failure. (BOOT_DEGRADED setting bogus, Bug #539597)

File a bug

Server with LUKS on RAID1 install

Case ID: SR1I-002

1. Boot CD and run the CD self-check (then reboot)

2. Select Install to hard disk

3. Choose your language, country and keyboard layout
4. Set hostname: default - ubuntu
5. Partition disks: Custom partition scheme.
   1. RAID1 array for /boot
   2. RAID1 array with LUKS on it for /
   3. RAID1 array for swap (should it get encrypted automatically?)
   4. RAID1 array with LUKS on it for /home
6. Select "Yes" to the "boot degraded?" question
7. Select your time zone and set the system clock to UTC
8. User account: enter username and password
9. No software selection.
10. Reboot and login.
11. Make sure that the root and home file systems are mounted from luks devices:

    • sudo mount

12. Make sure that the swap partition is mounted from a md device (and encrypted?):

    • sudo cat /proc/swaps

13. Make sure that the luks devices and /boot use md devices:

    • sudo dmsetup deps

14. Make sure that the raid arrays are working:

    • cat /proc/mdstat

15. Make sure that grub has been installed on both disks:

    • sudo apt-get install -y binutils
      for i in $(sudo mdadm -Q --detail $(df -P /boot | grep ^/dev/ | cut -d" " -f1) | grep " /dev/" | awk '{print $NF}' | sed -e 's/[0-9]$//'); do sudo dd if=$i bs=512 count=1 2>/dev/null | strings -a | grep -q GRUB && echo $i: ok || echo $i: FAIL; done

16. Make sure that the BOOT_DEGRADED setting is "TRUE" in /etc/initramfs-tools/conf.d/mdadm:

    • cat /etc/initramfs-tools/conf.d/mdadm

17. Test booting from a cold-degraded array:
    1. Poweroff the system
    2. Disconnect one of the disk (disk 2) - disk 1 connected, disk2 disconnected.
    3. Poweron the system
    4. Check that system boots correctly from the degraded RAID1 array on a single disk (note that you may have to wait up to 5 minutes for mdadm to time out and boot into degraded mode):

       • cat /proc/mdstat

    5. Poweroff the system
    6. Disconnect the first disk (disk1) and reconnect the second disk (disk2) - disk 1 disconnected, disk 2 connected. This results in booting the other half of the array, to see if this array segmentation is detected correctly afterwards. (see Bug #557429)
    7. Poweron the system.
    8. Check that system boots correctly from the degraded RAID1 array on a single disk:

       • cat /proc/mdstat

18. Test automatic re-syncing of degraded array:
    1. Poweroff the system
    2. Reconnect the first disk (so both are now connected)
    3. Poweron the system
19. Wait for both drives to be back in sync:

    • cat /proc/mdstat

20. Change "Do you want to boot degraded?" answer to "No":

    • sudo dpkg-reconfigure mdadm

21. Make sure that the BOOT_DEGRADED setting is "FALSE" in /etc/initramfs-tools/conf.d/mdadm:

    • cat /etc/initramfs-tools/conf.d/mdadm

22. Test booting from a cold-degraded array:
    1. Poweroff the system
    2. Disconnect one of the disk (disk 2) - disk 1 connected, disk2 disconnected.
    3. Poweron the system
    4. Check that on boot a question to enable and boot from a degraded array is asked.
    5. Say yes
    6. Check that system boots correctly from the degraded RAID1 array:

       • cat /proc/mdstat

    7. Poweroff the system
    8. Disconnect the first disk (disk1) and reconnect the second disk (disk2) - disk 1 disconnected, disk 2 connected.
    9. Poweron the system.
    10. Check that on boot a question to enable and boot from a degraded array is asked.
    11. Say yes
    12. Check that system boots correctly from the degraded RAID1 array:

        • cat /proc/mdstat

23. Re-add/sync the arrays again
    1. Poweroff the system
    2. Reconnect the first disk (so both are now connected)
    3. Poweron the system
    4. Add the missing drives back to the RAIDs:

       • sudo mdadm -a /dev/mdX /dev/MISSING-DEVICE

24. Test booting from a hot-degraded array:
    1. Remove (unplug/fail) one disk from the running system.
    2. Check if users/admin get a notification message and beep about the failing raid.
    3. Reboot, verify that system comes up degraded without failure. (BOOT_DEGRADED setting bogus, Bug #539597)

File a bug

Server with LVM on LUKS on RAID1 install

Case ID: SR1I-003

1. Boot CD and run the CD self-check (then reboot)

2. Select Install to hard disk

3. Choose your language, country and keyboard layout
4. Set hostname: default - ubuntu
5. Partition disks: Custom partition scheme.
   1. RAID1 array for /boot
   2. RAID1 array with LUKS on it, with LVM on it, for /, /swap and /home
6. Select "Yes" to the "boot degraded?" question
7. Select your time zone and set the system clock to UTC
8. User account: enter username and password
9. No software selection.
10. Reboot and login.
11. Make sure that the root and home file systems are mounted from mapper devices:

    • sudo mount

12. Make sure that the swap partition is mounted from a mapper device:

    • sudo cat /proc/swaps

13. Make sure that lvm uses luks, and luks is using a md device:

    • sudo dmsetup deps

14. Make sure that the raid arrays are working:

    • cat /proc/mdstat

15. Make sure that grub has been installed on both disks:

    • sudo apt-get install -y binutils
      for i in $(sudo mdadm -Q --detail $(df -P /boot | grep ^/dev/ | cut -d" " -f1) | grep " /dev/" | awk '{print $NF}' | sed -e 's/[0-9]$//'); do sudo dd if=$i bs=512 count=1 2>/dev/null | strings -a | grep -q GRUB && echo $i: ok || echo $i: FAIL; done

16. Make sure that the BOOT_DEGRADED setting is "TRUE" in /etc/initramfs-tools/conf.d/mdadm:

    • cat /etc/initramfs-tools/conf.d/mdadm

17. Test booting from a cold-degraded array:
    1. Poweroff the system
    2. Disconnect one of the disk (disk 2) - disk 1 connected, disk2 disconnected.
    3. Poweron the system
    4. Check that system boots correctly from the degraded RAID1 array on a single disk (note that you may have to wait up to 5 minutes for mdadm to time out and boot into degraded mode):

       • cat /proc/mdstat

    5. Poweroff the system
    6. Disconnect the first disk (disk1) and reconnect the second disk (disk2) - disk 1 disconnected, disk 2 connected.
    7. Poweron the system.
    8. Check that system boots correctly from the degraded RAID1 array on a single disk:

       • cat /proc/mdstat

18. Test automatic syncing of degraded array:
    1. Poweroff the system
    2. Reconnect the first disk (so both are now connected)
    3. Poweron the system
19. Wait for both drives to be back in sync:

    • cat /proc/mdstat

20. Change "Do you want to boot degraded?" answer to "No":

    • sudo dpkg-reconfigure mdadm

21. Make sure that the BOOT_DEGRADED setting is "FALSE" in /etc/initramfs-tools/conf.d/mdadm:

    • cat /etc/initramfs-tools/conf.d/mdadm

22. Test booting from a cold-degraded array:
    1. Poweroff the system
    2. Disconnect one of the disk (disk 2) - disk 1 connected, disk2 disconnected.
    3. Poweron the system
    4. Check that on boot a question to enable and boot from a degraded array is asked.
    5. Say yes
    6. Check that system boots correctly from the degraded RAID1 array:

       • cat /proc/mdstat

    7. Poweroff the system
    8. Disconnect the first disk (disk1) and reconnect the second disk (disk2) - disk 1 disconnected, disk 2 connected.
    9. Poweron the system.
    10. Check that on boot a question to enable and boot from a degraded array is asked.
    11. Say yes
    12. Check that system boots correctly from the degraded RAID1 array:

        • cat /proc/mdstat

23. Re-add/sync the arrays again
    1. Poweroff the system
    2. Reconnect the first disk (so both are now connected)
    3. Poweron the system
    4. Add the missing drives back to the RAIDs:

       • sudo mdadm -a /dev/mdX /dev/MISSING-DEVICE

24. Test booting from a hot-degraded array:
    1. Remove (unplug/fail) one disk from the running system.
    2. Check if users/admin get a notification message and beep about the failing raid.
    3. Reboot, verify that system comes up degraded without failure. (BOOT_DEGRADED setting bogus, Bug #539597)

File a bug