Oracle ZFS Storage Appliance: Renumbering of LUN IDs with the Openstack Cinder Driver (Doc ID 2119213.1)

Last updated on FEBRUARY 02, 2022

Applies to:

Symptoms

A LUN ID which is inserted into the (Cinder) mysql DB by OpenStack and which is put (by OpenStack) into the Zone configuration files differs - after a reboot of the entire OpenStack Environment - from the LUN ID offered by the ZFS-SA (it was created by OpenStack as well).

The mysql DB (Cinder) and the Zone-XML have still the original values, but the LUN in the ZFS_SA has got a new one.

In the ZS3 Audit-Log one can see that "Cinder" is Setting (a change) the assigned_number of the LUN to '-1'.

• OS Solaris 11.2 sru 10 with openstack Juno
• ZS3 Firmware 2013.06.05.4.2.1-1.1

The reboot of the OpenStack issues commands which assign the LUNs to an Initiator Group they already belonged before.

This causes the LUN to get a new ID assigned - leading to all kind of issues on the client side and for the applications/databases.

As the LUNs get new IDs assigned, the LUNs are not correctly assigned, while the configuration of OpenStack has not been changed for the remaining LUNs.

Steps to reproduce

1. Create LUNs with IDs 1 to 10 and add them to an initiator group.
2. Have everything configured in OpenStack and Database
3. Remove some of the LUNs from the appliance and configuration to create gaps
4. Boot the complete OpenStack on the Solaris Client

Changes

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.