This document provides answers for the following types of questions:
What are the roles of the quorum disk and quorum disk candidates?
How can the current quorum disk candidates be identified?
What are the recommendations for changing the managed disks assigned as quorum candidate disks?
An SVC cluster automatically assigns quorum disk candidates, but it is good practice to review the quorum disk assignments when adding new storage to a cluster or removing existing storage.
The cluster uses the quorum disk for two purposes: as a tie breaker in the event of a SAN fault, when exactly half of the nodes that were previously a member of the cluster are present, and to hold a copy of important cluster configuration data. Just over 256 MB is reserved for this purpose on each quorum disk candidate.
There is only one active quorum disk in a cluster; however, the cluster uses three managed disks as quorum disk candidates. The cluster automatically selects the actual active quorum disk from the pool of assigned quorum disk candidates. In SVC software version 5.1.0 and above, the active quorum disk can be specified with the svctask setquorum command and using the -active flag option.
The purpose of the other quorum disk candidates is to provide redundancy if a quorum disk fails prior to a cluster being partitioned. You should assign quorum disk candidates on multiple controllers to avoid the possibility of losing all the quorum disk candidates with a single failure.
Note: Mirrored VDisks may be taken offline if there is no quorum disk available. This behaviour occurs because the synchronisation status for mirrored VDisks is recorded on the quorum disk.
The following general guidance is provided when changing the managed disks that are assigned as quorum candidate disks:
- When possible, aim to distribute the quorum candidate disks so that each MDisk is provided by a different storage system. Refer to the Supported Hardware List for information on which storage systems are supported for quorum disk use.
- Ensure the status of the managed disk being assigned as a quorum candidate disk is reported as "online" before issuing the setquorum command.
- Allow at least two minutes between issuing consecutive setquorum commands. This delay allows time to ensure the quorum disk update is complete before another quorum disk change is made.
Additional guidance for split-cluster configurations:
To provide protection against failures that affect an entire location (for example a power failure), you can use VDisk mirroring in conjunction with a configuration that splits a single cluster between two physical locations. Example split-cluster configurations are available in the SAN Volume Controller Information Center.
Refer to the SVC Software Installation and Configuration Guide for information on the configuration requirements and restrictions for a supported split-cluster configuration. Contact your IBM regional advanced technical specialist for detailed guidance on configuring a split-cluster configuration for high availability purposes.
In the general case, where the nodes in a cluster have been split between sites, the SVC cluster should be configured such that:
- Site 1: Half of SVC cluster nodes + one quorum disk candidate
- Site 2: Half of SVC cluster nodes + one quorum disk candidate
- Site 3: Active quorum disk
This configuration ensures that there is always a quorum disk available, even after a single site failure.
SVC 5.1.0 and above
In SVC software version 5.1.0 and above the active quorum disk can be specified with the svctask setquorum command and using the -active flag option. The current quorum disk status can be viewed using the svcinfo lsquorum command
SVC 4.3.1 and earlier
The difficulty in implementing this configuration is that in SVC software versions prior to 5.1.0, it is not possible to identify which of the quorum disk candidates is the active quorum disk for the cluster. However, the following procedure can be used to define a preference for the quorum disk usage (prior to a failure or managed disk configuration change) and ensure the quorum disk candidate at site 3 will initially be assigned as the active quorum disk.
- Identify the 3 MDisks at different sites that are to be used as quorum disk candidates
- Make sure none of these MDisks are currently assigned as quorum disk candidates (see the guidance below for how to identify the current quorum disk candidates). You may need to temporarily assign other MDisks as quorum disk candidates to achieve this
- Assign the chosen MDisk at site 3 as quorum disk candidate index 0 (preferred quorum disk)
- Assign the chosen MDisk at site 2 as quorum disk candidate index 1
- Assign the chosen MDisk at site 1 as quorum disk candidate index 2
The quorum index (0, 1, 2) assigned to each of the 3 managed disks does not influence the quorum disk selection algorithm. However, by assigning the index values in this manner you have a record of the ordering. It is the order in which the MDisks are assigned as quorum disk candidates which defines the preference for initial use as the active quorum disk.
Example scenarios which result in changes to the active quorum disk:
- Site 3 is powered off, or connectivity to the site is broken
- Cluster chooses a quorum disk candidate at site 2 to become the active quorum disk
- Site 3 is powered on, or connectivity to the site is restored
- The SVC administrator must follow the procedure above and reassign all three quorum disks to ensure the active quorum disk is now located at site 3 again
- Controller hosting preferred quorum disk at site 3 is removed from the configuration
- Cluster automatically configures a new quorum disk candidate at site 1 or 2 (if possible)
- Cluster chooses a quorum disk candidate at site 1 or 2 to become the active quorum disk
- A new controller is added to site 3
- The SVC administrator must follow the procedure above and reassign all three quorum disks to ensure the active quorum disk is now located at site 3 again.
Identifying the current quorum disk candidates:
The following script can be used at the command-line interface to identify the current quorum candidate disks. Copy and paste this as a single command:
svcinfo lsmdisk -nohdr | while read id name status mode mdisk_grp_id mdisk_grp_name capacity ctrl_LUN controller_name mdisk_UID; do svcinfo lsmdisk $id | while read key value; do if [ "$key" == "quorum_index" ]; then if [ "$value" != "" ]; then echo "Quorum index $value : mdisk $id ($name), status=$status, controller=$controller_name"; fi; fi; done; done
Quorum index 0 : mdisk 0 (mdisk0), status=online, controller=controller0
Quorum index 1 : mdisk 2 (mdisk2), status=online, controller=controller1
Quorum index 2 : mdisk 15 (mdisk15), status=online, controller=controller2
Assigning quorum disk candidates:
Follow the links below for information on setting quorum candidate disks using the SVC Console (GUI) or command-line interface (CLI).
|Storage Virtualization||SAN Volume Controller||V4.2.x||Platform Independent||Version Independent|
|Storage Virtualization||SAN Volume Controller||V4.3.x||Platform Independent||Version Independent|