Unbind a Neo4j cluster server

Command

The cluster state of a Causal Cluster member can be removed by using the following command:

Syntax

neo4j-admin unbind [--verbose]
                   [--expand-commands]
                   [--archive-cluster-state=<true/false>]
                   [--archive-path=<path>]

Options

Option Default Description

--verbose

Enable verbose output.

--expand-commands

Allow command expansion in config value evaluation.

--archive-cluster-state

false

Enable or disable the cluster state archiving.

--archive-path

Destination (file or folder) of the cluster state archive.

Limitations

The Neo4j server process must be shut down before running the neo4j-admin unbind command.

Examples of usage

You can use the neo4j-admin unbind command to:

  • Turn a cluster member into a standalone server:

    To start the Neo4j server in single (standalone) mode after unbinding it from the cluster, first set dbms.mode=SINGLE in neo4j.conf.

  • Seed a Causal Cluster with existing store files:

    To seed a new cluster using the store files of another cluster, you must first run neo4j-admin unbind on each server. For more information about seeding Causal Clusters, see Seed a cluster.

    If a cluster holds a previous version of any of the databases being seeded, you must DROP those databases before seeding. Alternatively, you can stop every instance, unbind them from the cluster using neo4j-admin unbind and then forcefully restore the correct seeds (backups) for the databases in question. If you do not DROP or unbind before seeding, either with neo4j-admin restore or neo4j-admin load, the database’s store files and cluster state will be out of sync, potentially leading to logical corruptions.

  • Recover a Causal Cluster:

    In the event of serious failures you may need to recover an entire Causal Cluster from backups. Before restoring those backups, you must first run neo4j-admin unbind on each server. For more information about recovering databases from online backups, see Restore a database backup.

    From Neo4j version 4.0.0 onwards, you must run the neo4j-admin unbind command on both Read Replicas and Core members.

Archive cluster state

If something goes wrong and debugging is needed, you can archive the cluster state, from the <neo4j-home> folder, run the neo4j-admin server unbind command with the arguments --archive-cluster-state=true and --archive-path=<destination-folder>:

bin/neo4j-admin unbind --archive-path=/path/to/archive-folder --archive-cluster-state=true

The default resultant file is named:

unbound_cluster_state.<YYYYMMDDHH24MM>.zip