Migrate a two-node cluster to an HA setup
This document describes a guided procedure to migrate from a two-node cluster (nodes n0 and n1) to a two-node high availability (HA) setup. The procedure helps you preserve service continuity while you transition cluster-specific networking and configuration dependencies to an HA model.
Notes:
- These steps are primarily for a two-node cluster (n0 and n1) to two-node HA migration.
- The sequence of steps in this document is provided as guidance only. You might have to adjust it to fit your specific environment and requirements.
Before you begin
- Validate your topology requirements.
- Identify feature-specific impacts, such as CLAG, ECMP, and GSLB.
Procedure
-
Mark node n1 as passive.
- Optional: use the passive timeout feature to drain connections gracefully before making n1 passive. For more information, see Graceful shutdown of nodes.
-
Take a full backup.
- Back up
ns.confand ZebOS. -
From the cluster IP (CLIP) address, run the following command:
add system backup - For more information, see Backup and restore of cluster setup.
-
Back up spotted configuration by using the following command:
shell spottedconfig <nodeID>This command is supported in release 14.1 and later.
- Back up
-
If CLAG is configured, make the necessary topology-specific adjustments.
- If only one interface is used between a node and the upstream switch, remove the LACP configuration from the upstream switch.
- If multiple links are used from a node (for example, n1) to the upstream switch, convert cluster-level LAG to node-level LAG on n1.
- Ensure that the LACP key (
lacpKey) configured on the link aggregation (LA) channel is updated consistently on both nodes to match the upstream switch configuration.
-
If ECMP is used for traffic distribution, update the dynamic routing configuration to align with the target HA traffic flow.
For more information, see Using the Equal Cost Multiple Path (ECMP).
-
Remove cluster node n1 from the CLIP address.
-
Configure HA node n1 as secondary by setting the HA status as StaySecondary.
set ha node -hastatus STAYSECONDARY add ha node 1 <NSIP_of_n0> -
Apply the required configuration on node n1.
-
Apply the required extracted configuration and restructured spotted configuration taken in step 2.
Example: In a cluster setup, GSLB configuration such as site bindings might be associated with a node group. Identify and adapt such cluster-specific settings during migration.
-
For spotted configuration details, see Striped, partially striped, and spotted configurations.
-
-
Take a backup on n0 (same as step 2).
-
Disable the cluster instance from the NSIP address of node n0.
-
On node n1, enable HA using the following command:
set ha node -hastatus ENABLE -
Verify that the node n1 processes application traffic.
-
Remove the cluster instance on node n0 from NSIP.
-
Apply topology changes on node n0.
- Complete CLAG-related topology changes.
- Remove external traffic distribution mechanisms, if used (for example, ECMP), based on your topology plan.
-
Remove the backplane physically or disable backplane interfaces on both nodes.
- If necessary, convert the backplane to a HA SYNC VLAN. For more information, see Restricting high availability synchronization traffic to a VLAN.
-
On node n0, add n1 as the HA peer.
add ha node 1 <NSIP_of_n1> -
If necessary, force HA synchronization on node n0. Then apply the backed-up spotted configuration and any remaining required configuration.