Clusters

A cluster is a pair of nodes at a single site that share some configurations and provide automatic failover. An additional IP address is assigned as a Cluster Virtual IP address that can move between the nodes if failover occurs.

Certain settings such as network services and VPN settings can be configured for the cluster and these settings will override the individual node’s configuration.

Requirements

• Nodes in the cluster must be using the same model of appliances
• Nodes must be able to create direct TCP connections with each other
• Nodes must be at the same physical site.
  • Using multiple public cloud (AWS/Azure) availability zones (AZs) within the same region are acceptable as they provide highly reliable connectivity between the nodes.

Active Member Determination

The active member of a cluster is determined by the following factors:

• Cluster heartbeat communication
• Cluster mode
• Configured active member
• Cluster member health

Cluster Heartbeat Communication

Cluster members utilize a direct TCP connection to each other to determine if their partner is online and share their health status. Each node will listen on a configured heartbeat IP and port, while at the same time connecting to their partner’s configured heartbeat IP and port.

Heartbeat communication is configured on each node’s cluster page.

Cluster Mode

A cluster can be configured in two different modes to determine what happens when a failed member returns to healthy status:

• Automatic Failback (Default) - In this mode the member set as the Configured Active (see below) will maintain the active role as long as it is online and healthy.
• Manual Failback - In this mode, the active role only moves if either the current holder fails or the configured active is changed

Consider a cluster with members named Node1, the configured active, and Node2.

Cluster Timeout

The cluster will wait a configurable amount of time before considering a failed member as lost. This timeout is configurable on the cluster page.

Configured Active

Each cluster will have one configured or preferred active member. This is reflected in the overview section.

Change Configured Active

To change the configured active member:

1. Click the make active button in the row for the desired node.

   

2. A prompt will appear asking for confirmation. Click the “Confirm” button.

   

3. This change will be pushed to each member. They will then attempt to transfer the active role to the new configured active member assuming it is online and healthy. After this change the “Current Active” will be updated to reflect the new active member.

Cluster Member Health

There may be situations where both cluster members are online and can communicate with each other, but external conditions exist that make a node unsuitable to hold the active role. The Trustgrid node service monitors for such conditions and will make a node as unhealthy if one occurs. The node will release the active role and its standby member will take over if it is online and healthy.

When the condition clears the node will declare itself healthy and inform its partner member. Depending on the cluster mode it may reclaim the active role.

Cluster Member Health Conditions

• Loss of cluster heartbeat communication - If a node cannot communicate with its partner nodes on the configured IP and port it will declare that partner node unhealthy and claim the active role if it has not already.
• Interface Link (Up/Down) State - Any interface configured with an IP address in the Trustgrid is monitored for a successful connection to another network device
  • Example: In a two-interface setup it would be possible for the LAN interface to become unplugged but the node to remain online from the cloud and data plane perspective. However, the lack of LAN connectivity would prevent the node from delivering any Trustgrid services to that network
  • Note: If only a single node has an interface configured that is not critical for the node to function, such as an alternate MPLS path, you can enable the Ignore Health Check option on that interface to prevent it from marking the node as unhealthy if that interface goes down.
• Upstream Internet Issues - If a Trustgrid node is unable to build connections to both the Trustgrid control plane AND data plane connections to its gateways the node will be marked as unhealthy. This does require all the connections to be failing before it is triggered
  • Example: If an upstream internet provider or device experiences failure the node will not be able to provide any services.
• WAN Interface DHCP failure - If the WAN interface is configured to use DHCP and it does not receive a DHCP lease it will mark itself unhealthy.

Cluster Member Shared Configuration

Cluster members can share the configuration for the following services:

• Networking
  • Interfaces
    • Interface routes
    • Cluster VIP (supported in on-premise, traditional network environment)
    • AWS/Azure/GCP route table entries
  • VPN
  • Tunnels
  • VRFs
  • ZTNA gateway endpoints
  • Layer 4 services and connectors
• Compute
  • Container and command definitions
  • Resource limits

Tags

Tags are visible at the bottom of the overview page for the resource. Tags applied to a cluster will be applied to both members of the cluster.

Modifying Tags

To add a tag:

1. Click Add Tag.
2. A new row will appear at the bottom of the tags table. There will be a list of existing tag names for your organization. You can filter the list by typing in the field. You can either select an existing tag name, or create a new one by typing it out in full and then selecting New selection: tagName.

   

3. Next move to the values field. As with the name, existing values will be listed. To enter a new value type it in completely.
4. Click Save

Tag rows can be edited in-place. Change the name or value, then click Save.

To remove a tag, click the red X next to the tag name, then click Save.

Applying a Tag Filter to the Clusters Table

The clusters table can also be filtered to only show clusters with a specific tag name:value.

1. On the clusters table click Actions and select Add Tag Filter from the drop-down menu.

   

2. After clicking Add Tag Filter, select the tag-name field and you will see a list of tag-names available. Select the desired tag.

   

   You can also start typing to filter what tag names are shown.
3. Select the tag value field and you will see a list of available values. Select the desired value.
4. (Optional) Click Add Tag Filter to include an additional filter. Note that the two filters will be applied using AND only clusters with both tag name:value combinations matching will be shown.
5. Click Apply Tag Filter and the table will only show matching clusters.