Updated at July 17, 2026 02:08 PM

Backend store

Тarantool Cluster Manager uses an underlying data store (backend store) for its entities: users, roles, cluster connections, settings, and other objects that you manipulate in TCM. The backend store can be either an etcd or a Tarantool cluster.

For better reliability and scalability, the backend store works independently from TCM. For example, it can be the same ectd or Tarantool cluster that you use as a centralized configuration storage. This makes TCM stateless: all objects created or modified in its web UI are saved to the backend store, and nothing is stored inside the TCM instances themselves. Any number of instances can duplicate each other when connected to the same backend store. If you stop all instances, the store still contains their objects. You can continue working with them right after starting a new instance.

In addition to using an external backend store, you can run TCM with an embedded etcd or Tarantool instance to use as the backend store.

On this page, you will learn to connect TCM to backend stores of both types, or start TCM with an embedded backend store.

Setting up a backend store

The TCM backend store requires the same configuration as Tarantool centralized configuration storage. Follow the instructions in centralized_configuration_storage_set_up to set up a backend store.

Configuring backend store connection

The TCM's connection to its backend store is configured using the storage.* configuration options. The storage.provider option selects the store type. It can be either etcd or tarantool.

External etcd store

To use an etcd cluster as a TCM backend store, set the storage.provider option to etcd and specify connection parameters in storage.etcd.* options. A minimal etcd configuration includes the storage endpoints:


storage:  provider: etcd  etcd:    endpoints:      - http://127.0.0.1:2379

If authentication is enabled in etcd, specify storage.etcd.username and storage.etcd.password:


storage:  provider: etcd  etcd:    endpoints:      - http://127.0.0.1:2379    username: etcduser    password: secret

The TCM data is stored in etcd under the prefix specified in storage.etcd.prefix. By default, the prefix is /tcm. If you want to change it or store data of different TCM instances separately in one etcd cluster, set the prefix explicitly:


storage:  provider: etcd  etcd:    endpoints:      - http://127.0.0.1:2379    prefix: /tcm2

Other storage.etcd.* options configure various aspects of the etcd store connection, such as network timeouts and limits or TLS parameters. For the full list of the etcd TCM backend store options, see the TCM configuration reference.

External Tarantool-based store

To use a Tarantool cluster as a TCM backend store, set the storage.provider option to tarantool and specify connection parameters in storage.tarantool.* options. A minimal configuration includes the one or more addresses of the backend store instances:


storage:  provider: tarantool  tarantool:    addr: http://127.0.0.1:3301

Or:


storage:  provider: tarantool  tarantool:    addrs:      - http://127.0.0.1:3301      - http://127.0.0.1:3302      - http://127.0.0.1:3303

If authentication is enabled in the backend store, specify storage.tarantool.username and storage.tarantool.password:


storage:  provider: tarantool  tarantool:    addr: http://127.0.0.1:3301    username: tarantooluser    password: secret

The TCM data is stored in the Tarantool-based backend store under the prefix specified in storage.tarantool.prefix. By default, the prefix is /tcm. If you want to change it or store data of different TCM instances separately in one Tarantool cluster, set the prefix explicitly:


storage:  provider: tarantool  tarantool:    addr: http://127.0.0.1:3301    username: tarantooluser    password: secret    prefix: /tcm2

Other storage.tarantool.* options configure various aspects of TCM connection to the Tarantool-based backend store, such as network timeouts and limits or TLS parameters. For the full list of the Tarantool-based TCM backend store options, see the TCM configuration reference.

Embedded backend store

For development purposes, you can start TCM with an embedded backend store. This is useful for local runs when you don't have or don't need an external backend store.

An embedded TCM backend store is a single instance of etcd or Tarantool that is started automatically on the same host during the TCM startup. It runs in the background until TCM is stopped. The embedded backend store is persistent: if you start TCM again with the same backend store configuration, it restores the TCM data from the previous runs.

The embedded backend store parameters are configured using the storage.etcd.embed.* options for etcd or storage.tarantool.embed.* options for a Tarantool-based store.

To start TCM with an embedded etcd with default settings, set storage.etcd.embed.enabled to true and leave other storage.* options default:


storage.etcd.embed.enabled: true

You can use the following call to get TCM running with embedded etcd without a configuration file:


$ tcm --storage.etcd.embed.enabled

To start TCM with an embedded Tarantool storage with default settings:

set storage.provider to tarantool
set storage.tarantool.embed.enabled to true


storage:  provider: tarantool  tarantool.embed.enabled: true

With command-line arguments:


$ tcm --storage.provider=tarantool --storage.tarantool.embed.enabled

You can tune the embedded backend store, for example, enable and configure TLS on it or change its working directories or startup arguments. To set specific parameters, specify the corresponding storage.etcd.embed.* or storage.tarantool.embed.* options. For the full list of configuration options of embedded backend stores, see the TCM configuration reference.

Setting up a cluster of embedded backend stores

To simulate the production environment, you can form a distributed multi-instance cluster from embedded stores of multiple TCM instances. To do this, configure each TCM instance's embedded store to join each other.

For etcd, provide the embedded store clustering parameters storage.etcd.embed.* and specify the endpoints in storage.etcd.endpoints. The options that configure embedded etcd mostly match the etcd configuration options. For more information about these options, see the etcd documentation.

Below are example configurations of three TCM instances that start with embedded etcd instances and form an etcd cluster from them:

First instance:


http:  port: 8080storage:  provider: etcd  etcd:    endpoints:      - http://127.0.0.1:2379      - http://127.0.0.1:22379      - http://127.0.0.1:32379    embed:      enabled: true      name: infra1      endpoints:        - http://127.0.0.1:2379      advertises:        - http://127.0.0.1:2379      initial-cluster-state: new      initial-cluster: "infra1=http://127.0.0.1:12380,infra2=http://127.0.0.1:22380,infra3=http://127.0.0.1:32380"      initial-cluster-token: etcd-cluster-1      peer-endpoints:        - http://127.0.0.1:12380      peer-advertises:        - http://127.0.0.1:12380      workdir: node1.etcd

Second instance:


http:  port: 8081storage:  provider: etcd  etcd:    endpoints:      - http://127.0.0.1:2379      - http://127.0.0.1:22379      - http://127.0.0.1:32379    embed:      enabled: true      name: infra2      endpoints:        - http://127.0.0.1:22379      advertises:        - http://127.0.0.1:22379      initial-cluster-state: new      initial-cluster: "infra1=http://127.0.0.1:12380,infra2=http://127.0.0.1:22380,infra3=http://127.0.0.1:32380"      initial-cluster-token: etcd-cluster-1      peer-endpoints:        - http://127.0.0.1:22380      peer-advertises:        - http://127.0.0.1:22380      workdir: node2.etcd

Third instance:


http:  port: 8082storage:  provider: etcd  etcd:    endpoints:      - http://127.0.0.1:2379      - http://127.0.0.1:22379      - http://127.0.0.1:32379    embed:      enabled: true      name: infra3      endpoints:        - http://127.0.0.1:32379      advertises:        - http://127.0.0.1:32379      initial-cluster-state: new      initial-cluster: "infra1=http://127.0.0.1:12380,infra2=http://127.0.0.1:22380,infra3=http://127.0.0.1:32380"      initial-cluster-token: etcd-cluster-1      peer-endpoints:        - http://127.0.0.1:32380      peer-advertises:        - http://127.0.0.1:32380      workdir: node3.etcd

To set up a cluster from embedded Tarantool-based backend stores:

Specify the Tarantool cluster configuration in storage.tarantool.embed.config (as a plain text) or storage.tarantool.embed.config-file (as a YAML file).
Assign an instance name from this configuration to each instance using storage.tarantool.embed.args to each embedded store.

Below are example configurations of three TCM instances that start with embedded Tarantool-based backend stores and form a cluster from them: