Preparing Memory Machine to Work with Hazelcast
To prepare for taking snapshots of the Hazelcast cluster: install Memory Machine, enable namespace restore, and provide the Hazelcast credentials to all hosts in the cluster.
About this task
Repeat the following steps on every host in a Hazelcast cluster to prepare to capture a snapshot of the cluster.
Procedure
- Install Memory Machine on the host. See Installing MemVerge Memory Machine.
-
Start the mvcsnapd service:
# systemctl start mvcsnapd #
-
Configure the mvcsnapd service to start on system
boot:
# systemctl enable mvcsnapd
-
Enable the namespace restore feature in managed applications by adding a line
to the Memory Machine application configuration file:
# echo 'DisableCloneInstance: true' >> /etc/memverge/mvmalloc.yml #
Important: Enabling the namespace restore feature disables the Memory Machine clone feature on the host. See Cloning an Instance with CLI and Cloning an Instance with Memory Machine Dashboard. - Stop any application instances running in Memory Machine. See Stopping an Instance.
-
Restart the Memory Machine
DPME:
# systemctl restart mvmallocd #
-
Memory Machine uses the mvm_cluster_snapshot_hazelcast.sh script to
communicate with the Hazelcast cluster. Set the cluster credentials in this file
to allow Memory Machine to change the cluster state while taking a snapshot.
In the sbin subdirectory of the Memory Machine installation directory (the default is /opt/memverge/sbin), edit the mvm_cluster_snapshot_hazelcast.sh file to change the CLUSTER_NAME and CLUSTER_PASSWORD parameters to those for your cluster.
# User need to provide these credentials. They can modify this script # to retrieve the credentials in their preferred way. CLUSTER_NAME="your_cluster" CLUSTER_PASSWORD="your_cluster_password"