Red Hat GLOBAL FILE SYSTEM 5.0 Instrukcja Użytkownika Pobierz pdf (Strona 10)

Deploying a Failover Cluster

Configuring Shared Storage for Failover Clusters

Once you install Parallels Server Bare Metal 5.0, you need to configure the shared storage to use it

as the /vz partition. The space requirements for this partition are calculated on the basis of space

demands for virtual machines and Containers. The number of partitions must correspond to the

number of clustered services and be less than the number of physical servers in the cluster.

Make sure that all servers can access all shared Logical Unit Numbers (LUNs). The cluster will

decide itself where to start the clustered service; so it must be able to mount the partition with all

required virtual machines and Containers.

Using volume labels for partitions, you can greatly simplify the further management of your cluster.

To make things more easier, you can have a match between a clustered service and a partition

name. For example, you can make the PSBM-1 service use a partition with the psbm-1 label.

When working with volume labels, note that the volume labels do not correspond to mount points

which are always /vz on all servers in the cluster.

Note: If you are planning to use iSCSI storages, see first Configuring iSCSI Data Storages (p. 12) for

information on setting up this kind of storages. Once you are done, proceed with the instructions below.

Once you create all necessary LUNs, you need to create and configure partitions on them.

Creating Partitions on LUNs

First, create partitions on LUNs:

1 Make partitions on LUNs using the fdisk utility or other disk partitioning software.

2 Create the file system on the new partitions. For example, to create the ext3 file system on the

/dev/sdc1 partition and set the psbm-1 label for the clustered service, you can run this

command:

# mkfs.ext3 -L psbm-1 /dev/sdc1

3 Check that all shared partitions can be successfully mounted to each cluster node. To do this,

create a temporary partition on each node and mount the shared partitions, one after another,

to this temporary partition. For example, you can log in to each cluster node and use the

following commands to check that the partition with the psbm-1 label can be mounted to the

/vz_temp partition:

# mkdir /vz_temp

# mount LABEL=psbm-1 /vz_temp

Once you check that the shared partition has been successfully mounted, unmount it:

# umount /vz_temp

1 2 ... 5 6 7 8 9 10 11 12 13 14 15 ... 27 28

Brak uwag

Red Hat GLOBAL FILE SYSTEM 5.0 Instrukcja Użytkownika Strona 10