User`s guide

Upgrading Nodes in Clusters

Upgrading Nodes in a Data Sharing Cluster

Like it is the case with a failover cluster, the process of upgrading nodes in a data sharing cluster

also includes two steps:

1 Upgrading active nodes in the cluster.

2 Upgrading passive nodes in the cluster.

However, each of these steps contains additional operations that you need to perform to upgrade

cluster nodes.

Important!

1. Upgrade one cluster node at a time. Upgrading more than one node simultaneously is not

supported and may damage your system.

2. Due to discontinued support in Red Hat Enterprise Linux 6, Parallels Virtuozzo Containers 4.7 no

longer supports keeping /vz partitions on GFS1 volumes.

Checking Clustered Services

Each active node must be upgraded together with the clustered service running on it. So before

starting the upgrade procedure, make sure that you know which clustered service is running on

which node. To see the list of clustered services and the nodes where they are running, run this

command on any cluster node:

# clustat

Service Name Owner (Last) State

------- ---- ----- ------ -----

service:vz-0 server1.com started

service:vz-1 server2.com started

Once you upgrade all active nodes in the cluster, no clustered services must be left un-upgraded.

Upgrading Active Nodes in the Cluster

First, you need to upgrade all active nodes in the data sharing cluster. To upgrade an active node:

1 Freeze the clustered service running on all active nodes in the cluster. To do this, run the

following command on each active node:

# clusvcadm -Z clustered_service_name

(where clustered_service_name is the name of the clustered service)

If your version of clusvcadm does not support the -Z option, use the -l option instead. This

option prevents the specified clustered service from running on a local node (that is, on the

node where you run the command). If you use the -l option, execute the command on all

cluster nodes (both active and passive), except for the node you are upgrading.