Upgrading Unity -> help with RAID

Answered Question
Jun 9th, 2008

I am preparing to upgrade our Unity 4.X server to Unity 5.X. I am going to remove one of rhe RAID discs before starting the upgrade so that if the upgrade fails I will be able to roll back.

My question is; in a mirror array, which drive should be removed for backup? How do I safely rebuild the array if the upgrade is successful?

-Shikamaru

I have this problem too.
0 votes
Correct Answer by Michael Owuor about 8 years 6 months ago

Hi,

Here's the official stance regarding removing drives from the Unity server (you'll find this in all Release Notes up to the latest version 7.0 of Unity):

* Replacing Disks in a RAID

For any server in a Cisco Unity system (a Cisco Unity server, a failover server, a voice-recognition server, a Cisco Unity Bridge server, a message store server, or a dedicated DC/GC), Cisco Unity supports only replacing a defective disk in a RAID with a blank disk to repair the RAID. Replacing disks in a RAID for any other reason is not supported.

Caution Do not replace a disk in a RAID with a disk that contains data, even if the replacement disk was originally a disk in the same RAID in the same server.

http://www.cisco.com/en/US/docs/voice_ip_comm/unity/7x/release/notes/702curelnotes.html#wp49156

Recommendations for backing up Unity are documented in the Maintenance Guides for Unity:

http://www.cisco.com/en/US/docs/voice_ip_comm/unity/5x/maintenance/guide/ex/5xcumge045.html

Having said that, it is not uncommon for users in the field (including myself) to employ the method you describe of removing a drive (from slot 0) that is part of a pair of a RAID-1 array, performing some maintenance, and then later reinserting the drive and rebuilding the array, or reverting to the original configuration in the event of a problem. On servers with 4 drives like 7845's using RAID-1, people often pull a single drive from each RAID1 mirror set (from slot 0 and slot 2).

For CCM, the procedures for removing drives as a means of being able to quickly restore a system to its last good working configuration in the event of a problem during an upgrade or patch, are documented in several places. Here are a couple:

* Recover from Upgrade Failure on Cisco CallManager

http://www.cisco.com/warp/public/788/AVVID/disk_redundancy_mcs_9229.html#upgrade

* Reconfiguring If You Removed a Drive Before the Upgrade

http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/upgrade/4_2_1/revert.html#wp1040178

For Unity, if you chose to use this method, be careful to verify the type of RAID in use before pulling drives. Also make sure to backup your Unity server using DiRT as well as any other supported backup software as described in the Maintenance Guide above. After all, HP guys will tell you that technically, RAID is a means of achieving fault tolerance, and is not intended to be an application backup and restore mechanism.

Hope this helps.

Regards,

Michael.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (1 ratings)
Loading.
egravel Mon, 06/09/2008 - 17:07

If it is a RAID 1 array, anyone of the 2 disc can be removed. If it is RAID 5, with 3 disks, you can not remove a disk.

To rebuild the array simply put back the disk in the same slot with the server running and it should rebuild by it self.

shikamarunara Mon, 06/09/2008 - 17:10

egravel,

This is the part that confuses me. If I take out the second disc and do the upgrade on the first disc, how does the system know which disc to rebuild with, since both of them would be different after the upgrade process?

-Shikamaru

Correct Answer
Michael Owuor Mon, 06/09/2008 - 22:27

Hi,

Here's the official stance regarding removing drives from the Unity server (you'll find this in all Release Notes up to the latest version 7.0 of Unity):

* Replacing Disks in a RAID

For any server in a Cisco Unity system (a Cisco Unity server, a failover server, a voice-recognition server, a Cisco Unity Bridge server, a message store server, or a dedicated DC/GC), Cisco Unity supports only replacing a defective disk in a RAID with a blank disk to repair the RAID. Replacing disks in a RAID for any other reason is not supported.

Caution Do not replace a disk in a RAID with a disk that contains data, even if the replacement disk was originally a disk in the same RAID in the same server.

http://www.cisco.com/en/US/docs/voice_ip_comm/unity/7x/release/notes/702curelnotes.html#wp49156

Recommendations for backing up Unity are documented in the Maintenance Guides for Unity:

http://www.cisco.com/en/US/docs/voice_ip_comm/unity/5x/maintenance/guide/ex/5xcumge045.html

Having said that, it is not uncommon for users in the field (including myself) to employ the method you describe of removing a drive (from slot 0) that is part of a pair of a RAID-1 array, performing some maintenance, and then later reinserting the drive and rebuilding the array, or reverting to the original configuration in the event of a problem. On servers with 4 drives like 7845's using RAID-1, people often pull a single drive from each RAID1 mirror set (from slot 0 and slot 2).

For CCM, the procedures for removing drives as a means of being able to quickly restore a system to its last good working configuration in the event of a problem during an upgrade or patch, are documented in several places. Here are a couple:

* Recover from Upgrade Failure on Cisco CallManager

http://www.cisco.com/warp/public/788/AVVID/disk_redundancy_mcs_9229.html#upgrade

* Reconfiguring If You Removed a Drive Before the Upgrade

http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/upgrade/4_2_1/revert.html#wp1040178

For Unity, if you chose to use this method, be careful to verify the type of RAID in use before pulling drives. Also make sure to backup your Unity server using DiRT as well as any other supported backup software as described in the Maintenance Guide above. After all, HP guys will tell you that technically, RAID is a means of achieving fault tolerance, and is not intended to be an application backup and restore mechanism.

Hope this helps.

Regards,

Michael.

Actions

This Discussion