12.5. Upgrading an HA Cluster

Home Prev	cOS Core 14.00.17 Administration Guide	Next

The cOS Core software versions running on the master and slave in an HA cluster should be the same. When a new cOS Core version becomes available and is to be installed on both units, the upgrade is performed on one unit at a time as though they were standalone units.

The central principle in the upgrade process for a cluster is that upgrading the inactive unit will not affect the operation of the cluster and only momentarily make the inactive unit unavailable.

The overall sequence of steps to follow is:

Identify which unit is inactive and upgrade that first.
When the inactive unit is once again synchronized with the active unit, cause a failover to occur so that the inactive becomes the active unit.
Now upgrade the inactive unit. Both units will then resynchronize and have the new cOS Core version.

These three steps will now be broken down into a more detailed description:

A. Establish which is the inactive unit in the cluster

The currently inactive unit will be upgraded first so it is necessary to identify this. To do this, connect with a CLI console to one of the cluster units and issue the ha command. The typical output if the unit is active is shown below.

Device:/> ha
		
This device is a HA SLAVE
This device is currently ACTIVE (will forward traffic)
This device has been active: 430697 sec
HA cluster peer is ALIVE

This unit (the slave) is the currently active unit, so the other one (the master) is the inactive unit.

B. Upgrade the inactive unit

Once the inactive unit is identified, upgrade this unit with the new cOS Core version. This is done exactly as though the unit were not in a cluster. For example, the Web Interface can be used to do the upgrade.

Important: Make sure the inactive unit is ALIVE

	Important: Make sure the inactive unit is ALIVE
Before going to the next step make sure the inactive unit is fully operational and synchronized with the active unit after the software upgrade completes. To do this, issue the CLI ha command on the inactive unit. The output from the command should indicate that the status is ALIVE. `Device:/>` ha This device is a HA SLAVE This device is currently INACTIVE (won't forward traffic) This device has been inactive: 2 sec HA cluster peer is ALIVE

Before going to the next step make sure the inactive unit is fully operational and synchronized with the active unit after the software upgrade completes.

To do this, issue the CLI ha command on the inactive unit. The output from the command should indicate that the status is ALIVE.

Device:/> ha
			
This device is a HA SLAVE
This device is currently INACTIVE (won't forward traffic)
This device has been inactive: 2 sec
HA cluster peer is ALIVE

C. Cause a failover to occur

Now, connect to the active unit (which is still running the old cOS Core version) with a CLI console and issue the ha -deactivate command. This will cause the active unit to become inactive, and the inactive to become active.

Device:/> ha -deactivate
			
HA Was: ACTIVE
HA going INACTIVE...

To check that the failover has completed successfully, an ha command can be issued again and the text "INACTIVE" and "is ALIVE" should appear in the output.

D. Upgrade the newly inactive unit

When the failover is complete, upgrade the newly inactive unit with the new cOS Core version. Just like step B, this is done in the normal way as though the unit were not part of a cluster.

E. Wait for resynchronization

Once the second software upgrade is complete, two units will automatically resynchronize and the cluster will continue operation. The roles of active and inactive unit will have been reversed.

If it is desirable to make the active unit inactive, and the inactive unit active, the CLI command ha -activate can be entered on the inactive unit.