How to update Veritas Volume Manager (VxVM) UDID content for NDM migrated EMC SYMDEVs using vxdisk -c updateudid

Description
 

This article outlines how to update Veritas Volume Manager (VxVM) UDID content for NDM migrated EMC SYMDEVs using the vxdisk -c updateudid command.

Unique Disk Identifier (UDID)

Veritas Volume Manager (VxVM) refers to two UDID values, the UDID value returned by the Device Discovery Layer (DDL) which is the correct (true) UDID value of the LUN, whereas the UDID written on-disk can be from a different source.

The UDID structure consists of the Vendor ID (VID), Product ID (PID), Cabinet Serial Number (Cab_Serial_No) and Lun Serial Number (Lun_Serial_No - LSN) :

"VID , '_', PID , '_', CAB _Serial_No , '_', Lun_Serial_No "

As a result of clearing/resetting the spoofing information, this leads to the expected VxVM udid_mismatch flag being set, as the true UUID in the kernel differs to what is written on-disk.
 

DELL/EMC NDM (Non-Disruptive Migration) overview

The DELL/EMC NDM solution is designed to help automate the process of migrating host applications to a PowerMax, VMAX All-Flash, or VMAX3 enterprise storage array with no downtime.

The NDM solution leverages VMAX SRDF replication technologies to move the application data from the source EMC storage array to the new target EMC storage array. 

Once the NDM array migration is complete, the process can result in stale historical information being reported to Veritas Dynamic Multi-Pathing (DMP).

To resolve the stale LUN information, the spoofing identity needs to be reset for the NDM migrated EMC SYMDEVs.

This can then lead to Veritas Volume Manager (VxVM) reporting the udid_mismatch flag once the SYMDEV identity has been reset.

Manipulation of device IDs and host paths
 

Two of the underlying processes that ensure that NDM is non-disruptive is the technologies ability to maintain device visibility at all times by spoofing and swapping devices IDs between source and target devices.

NDM is able to migrate data and cut over to the target array non-disruptively by both swapping device IDs between the source and target devices and manipulating the paths from the host to both arrays. 

The device ID contains the device’s unique WWN and other information about it, such as a device identifier that the user has assigned to a device through Solutions Enabler or Unisphere. 

All of this information is copied to the target devices.

NDM performs the data migration and device ID swap without the host being aware. 

The path management changes appear as either the addition of paths or the removal of paths to the existing source device. 

To the host and application, there is no change in the device that it is accessing and access to the device is maintained throughout the entire migration process.
 

Steps:

PLAN A

Upgrade InfoScale to 7.4.2.x or higher followed by a reboot of the upgrade host.

1. Disable Veritas Volume Manager (VxVM) from auto-starting:

# touch /etc/vx/reconfig.d/state.d/install-db

2. InfoScale 7.4.2.x Upgrade Complete.

3. Ensure no VxVM disk groups are deported and vxconfigd is "not-running".

# vxdg list

# vxdctl stop

# vxdctl mode

4. Storage Team reset spoof information for EMC NDM migrated devices

# symdev -sid SID reset -identity SYMDEV -noprompt

5. Reboot the system

6. Check for the udid_mismatch.

# vxdisk -eo alldgs list | grep udid

7. Update UDID for NDM related disks reporting udid_mismatch

# vxdisk list | grep udid_mismatch | awk '{print "vxdisk -c updateudid", $1}'

Note: Check the disks listed are the correct ones, before executing:

# vxdisk list | grep udid_mismatch | awk '{print "vxdisk -c updateudid", $1}' | sh -x

If in any doubt, contact Veritas Technical Support.
 

The UDID information can be updated to make it appear as a STANDARD (STD) disk once again, using the command:

# vxdisk -c updateudid veritas-disk-access-name


8. Check the revised status for the disks and import required disk groups

# vxdisk -eo alldgs list
# vxdisk -o udid list

# vxdg import dgname

Note: if VCS is configured and managing the VxVM disk groups, deport disk group(s)

# vxdg deport dgname

9.] Remove /etc/vx/reconfig.d/state.d/install-db file
 

10.] Starts Veritas Cluster Server (VCS) as required.



PLAN B

Manual steps used in place of rebooting host following InfoScale to 7.4.2.x or higher upgrade. Consult Veritas Technical Support for guidance if performing the following command sequence.


Upgrade InfoScale to 7.4.2.x or higher followed by a reboot of the upgrade host.

1.  Disable Veritas Volume Manager (VxVM) from auto-starting:

# touch /etc/vx/reconfig.d/state.d/install-db

2.  InfoScale 7.4.2.x Upgrade Complete.

3. Start VxVM in a disabled state

# vxconfigd -d
# vxdctl init

4. Check for the excluded devices.

# vxddladm list devices
 

5. Ensure no VxVM disk groups are imported and vxconfigd is "not-running".

# vxdg list

# vxdctl stop

# vxdctl mode

6. Storage Team reset spoof information for EMC NDM migrated devices

# symdev -sid SID reset -identity SYMDEV -noprompt


7. Once complete, start VxVM in a disabled state

# vxconfigd -d
# vxdctl init

8. Remove the devices that are in the error state.

# vxdisk -q list | grep error | awk '{print "vxdisk rm ", $1}' | sh -x

9. Exclude all devices from DMP control

# vxdmpadm exclude all

10.  Make a copy of the current /etc/vx/array.info file

# mv /etc/vx/array.info /etc/vx/array.info.save

11. Remove DDL & DMP customized configuration settings and reset enclosure naming & reindexing

# vxddladm -c assign names

Note: Customized DMPNODE & enclosure names will be removed with the "-c" option above.


12. Include and rediscover all disks under DMP

# vxdmpadm include all

13. Restart & reset vxconfigd

# vxconfigd -k -r reset
 

14. Check for the udid_mismatch flag

# vxdisk -eo alldgs list | grep udid

15. Update UDID for NDM related disks reporting udid_mismatch

# vxdisk list | grep udid_mismatch | awk '{print "vxdisk -c updateudid", $1}'

Note: Check the disks listed are the correct ones, before executing:

# vxdisk list | grep udid_mismatch | awk '{print "vxdisk -c updateudid", $1}' | sudo sh -x

16. Check the revised disk information and import VxVM disk groups as required

# vxdisk -eo alldgs list

# vxdisk -o udid list

17. Remove /etc/vx/reconfig.d/state.d/install-db file

19. Starts Veritas Cluster Server (VCS) as required.

Note: When possible reboot the server.