How to manage the cluster

  • 3 minute(s) read
Prev Next

The entire process of configuring a cluster and checking its status can be performed using the cluster command. In addition to the process, we will use the application command to start instances and define a primary member.


mt4adm@vmdf-giskard:~$ sudo orbit cluster --help
Usage: orbit cluster <command>

High Availability and Disaster Recovery settings tools

Commands:
cluster config
cluster start
cluster stop
cluster restart
cluster status
cluster arbiter
cluster net-check
cluster nodename
cluster rotate-ssl

Flags:
-h, --help Show context-sensitive help.

Create the cluster

Info
  • If there are doubts about enabling connections between the nodes, use the orbit cluster net-check command, adding the desired hosts for verification. More information in Firewall rules for the Segura® Platform.
  • Activate member A for both activation license and application activation.

Before configuring the cluster, it is crucial to include the cluster members in the allow list of the firewall of each member. For example, on member A, it is necessary to add member B to the allow list, and vice versa. More details in How to manage the firewall.

Attention

Having an arbitrator in a cluster is crucial for resolving disputes between nodes, ensuring system stability and integrity. It acts as a neutral third party in case of failure, avoiding inconsistencies and maintaining environment reliability. More information in Introduction Arbitrator.

To create the cluster, at least two members are required. For example, we will call the members A and B. See the following sections to learn how to configure the members.

Member A

  1. Run the sudo orbit cluster config command and configure the cluster on member A by filling the --ip parameter with the IP of member A and the --nodes parameter with the IPs of members A and B respectively.

mt4adm@vmdf-giskard:~$ sudo orbit cluster config
--ip=172.18.77.184
--nodes=172.18.77.184,172.18.77.186
--segment=0
--latency="low"

Are you sure you want to proceed: y
Done!
No errors reported
  1. Run the orbit cluster start command and wait at least 1 minute for normalization.

mt4adm@vmdf-giskard:~$ sudo orbit cluster start
Are you sure you want to proceed: y
Done!
No errors reported
Attention

For clusters with members in different data centers, it is recommended to use high latency between the members.

Info

Members from the same data center should be configured with the same segment. If a data center has only one member, you can leave the Network Segments field empty.

Member B

  1. Run the sudo orbit cluster config command and configure the cluster on member B by filling the --ip parameter with the IP of member B and the --nodes parameter with the IPs of members A and B respectively.

mt4adm@vmdf-giskard:~$ sudo orbit cluster config
    --ip=172.18.77.186
    --nodes=172.18.77.184,172.18.77.186
    --segment=1
    --latency="low"

Are you sure you want to proceed: y
Done!
No errors reported
  1. Run the orbit cluster start command and wait at least 1 minute for normalization.

mt4adm@vmdf-giskard:~$ sudo orbit cluster start
Are you sure you want to proceed: y
Done!
No errors reported
Attention

To ensure proper cluster operation, it is essential to maintain network latency between nodes below 30ms. Additionally, specific recovery settings are required to facilitate recovery in case of failure. You can find details on these settings at the How to enable Recovery.

After both members are successfully configured, go back to the member A and make it primary through the sudo orbit application master command.


mt4adm@vmdf-giskard:~$ sudo orbit application master
The application services will be stopped or restarted during the process.
Are you sure you want to proceed: y

Application: Active
Replication: Inactive
Instance:    Primary

Cluster status

Through the orbit cluster status command, you can observe various properties of the cluster. Since there is a lot of information, we will not list all the details in the manual. But we will present the essential information to understand if the cluster is active and without problems.

Pay attention to the final block of the status, called Cluster nodes. In this block, you will see who the cluster members are and the synchronization timestamp between them. There is also the status of the current member which should be marked as synced.


mt4adm@vmdf-giskard:~$ sudo orbit cluster status

...

============================================================

Cluster nodes

Cluster member: ID [0] - UUID [64661644-b0df-11ea-80b2-8ee23c1303c0] -
    Hostname [vmdf-giskard-3232290344] - Timestamp [1592428528]
Cluster member: ID [1] - UUID [9dff00d8-b0df-11ea-86c4-83725d654e03] -
    Hostname [vmdf-giskard-3232290348] - Timestamp [1592428528]

Cluster UUID: 64679b9f-b0df-11ea-a6e5-67ba900fecc0

vmdf-giskard details: status=synced
vmdf-giskard is primary node ?: 1

============================================================