Deploy a TiDB Cluster Using TiUP

TiUP is a cluster operation and maintenance tool introduced in TiDB 4.0. TiUP provides TiUP cluster, a cluster management component written in Golang. By using TiUP cluster, you can easily perform daily database operations, including deploying, starting, stopping, destroying, scaling, and upgrading a TiDB cluster, and manage TiDB cluster parameters.

TiUP supports deploying TiDB, TiFlash, TiDB Binlog, TiCDC, and the monitoring system. This document introduces how to deploy TiDB clusters of different topologies.

Step 1. Prerequisites and precheck

Make sure that you have read the following documents:

Step 2. Deploy TiUP on the control machine

You can deploy TiUP on the control machine in either of the two ways: online deployment and offline deployment.

Deploy TiUP online

Log in to the control machine using a regular user account (take the tidb user as an example). Subsequent TiUP installation and cluster management can be performed by the tidb user.

  1. Install TiUP by running the following command:

    curl --proto '=https' --tlsv1.2 -sSf https://tiup-mirrors.pingcap.com/install.sh | sh
  2. Set TiUP environment variables:

    1. Redeclare the global environment variables:

      source .bash_profile
    2. Confirm whether TiUP is installed:

      which tiup
  3. Install the TiUP cluster component:

    tiup cluster
  4. If TiUP is already installed, update the TiUP cluster component to the latest version:

    tiup update --self && tiup update cluster

    If “Update successfully!” is displayed, the TiUP cluster is updated successfully.

  5. Verify the current version of your TiUP cluster:

    tiup --binary cluster

Deploy TiUP offline

Perform the following steps in this section to deploy a TiDB cluster offline using TiUP:

Prepare the TiUP offline component package

To prepare the TiUP offline component package, you can manually pack an offline component package using tiup mirror clone.

  1. Install the TiUP package manager online.

    1. Install the TiUP tool:

      curl --proto '=https' --tlsv1.2 -sSf https://tiup-mirrors.pingcap.com/install.sh | sh
    2. Redeclare the global environment variables:

      source .bash_profile
    3. Confirm whether TiUP is installed:

      which tiup
  2. Pull the mirror using TiUP.

    1. Pull the needed components on a machine that has access to the Internet:

      tiup mirror clone tidb-community-server-${version}-linux-amd64 ${version} --os=linux --arch=amd64

      The command above creates a directory named tidb-community-server-${version}-linux-amd64 in the current directory, which contains the component package necessary for starting a cluster.

    2. Pack the component package by using the tar command and send the package to the control machine in the isolated environment:

      tar czvf tidb-community-server-${version}-linux-amd64.tar.gz tidb-community-server-${version}-linux-amd64

      tidb-community-server-${version}-linux-amd64.tar.gz is an independent offline environment package.

  3. Customize the offline mirror, or adjust the contents of an existing offline mirror.

    If you want to adjust an existing offline mirror (such as adding a new version of a component), take the following steps:

    1. When pulling an offline mirror, you can get an incomplete offline mirror by specifying specific information via parameters, such as the component and version information. For example, you can pull an offline mirror that includes only the offline mirror of TiUP v1.11.0 and TiUP Cluster v1.11.0 by running the following command:

      tiup mirror clone tiup-custom-mirror-v1.11.0 --tiup v1.11.0 --cluster v1.11.0

      If you only need the components for a particular platform, you can specify them using the --os or --arch parameters.

    2. Refer to the step 2 of "Pull the mirror using TiUP", and send this incomplete offline mirror to the control machine in the isolated environment.

    3. Check the path of the current offline mirror on the control machine in the isolated environment. If your TiUP tool is of a recent version, you can get the current mirror address by running the following command:

      tiup mirror show

      If the output of the above command indicates that the show command does not exist, you might be using an older version of TiUP. In this case, you can get the current mirror address from $HOME/.tiup/tiup.toml. Record this mirror address. In the following steps, ${base_mirror} is used to refer to this address.

    4. Merge an incomplete offline mirror into an existing offline mirror:

      First, copy the keys directory in the current offline mirror to the $HOME/.tiup directory:

      cp -r ${base_mirror}/keys $HOME/.tiup/

      Then use the TiUP command to merge the incomplete offline mirror into the mirror in use:

      tiup mirror merge tiup-custom-mirror-v1.11.0
    5. When the above steps are completed, check the result by running the tiup list command. In this document's example, the outputs of both tiup list tiup and tiup list cluster show that the corresponding components of v1.11.0 are available.

Deploy the offline TiUP component

After sending the package to the control machine of the target cluster, install the TiUP component by running the following commands:

tar xzvf tidb-community-server-${version}-linux-amd64.tar.gz && \ sh tidb-community-server-${version}-linux-amd64/local_install.sh && \ source /home/tidb/.bash_profile

The local_install.sh script automatically runs the tiup mirror set tidb-community-server-${version}-linux-amd64 command to set the current mirror address to tidb-community-server-${version}-linux-amd64.

To switch the mirror to another directory, run the tiup mirror set <mirror-dir> command. To switch the mirror to the online environment, run the tiup mirror set https://tiup-mirrors.pingcap.com command.

Step 3. Initialize cluster topology file

Run the following command to create a cluster topology file:

tiup cluster template > topology.yaml

In the following two common scenarios, you can generate recommended topology templates by running commands:

  • For hybrid deployment: Multiple instances are deployed on a single machine. For details, see Hybrid Deployment Topology.

    tiup cluster template --full > topology.yaml
  • For geo-distributed deployment: TiDB clusters are deployed in geographically distributed data centers. For details, see Geo-Distributed Deployment Topology.

    tiup cluster template --multi-dc > topology.yaml

Run vi topology.yaml to see the configuration file content:

global: user: "tidb" ssh_port: 22 deploy_dir: "/tidb-deploy" data_dir: "/tidb-data" server_configs: {} pd_servers: - host: 10.0.1.4 - host: 10.0.1.5 - host: 10.0.1.6 tidb_servers: - host: 10.0.1.7 - host: 10.0.1.8 - host: 10.0.1.9 tikv_servers: - host: 10.0.1.1 - host: 10.0.1.2 - host: 10.0.1.3 monitoring_servers: - host: 10.0.1.4 grafana_servers: - host: 10.0.1.4 alertmanager_servers: - host: 10.0.1.4

The following examples cover seven common scenarios. You need to modify the configuration file (named topology.yaml) according to the topology description and templates in the corresponding links. For other scenarios, edit the configuration template accordingly.

ApplicationConfiguration taskConfiguration file templateTopology description
OLTPDeploy minimal topologySimple minimal configuration template
Full minimal configuration template
This is the basic cluster topology, including tidb-server, tikv-server, and pd-server.
HTAPDeploy the TiFlash topologySimple TiFlash configuration template
Full TiFlash configuration template
This is to deploy TiFlash along with the minimal cluster topology. TiFlash is a columnar storage engine, and gradually becomes a standard cluster topology.
Replicate incremental data using TiCDCDeploy the TiCDC topologySimple TiCDC configuration template
Full TiCDC configuration template
This is to deploy TiCDC along with the minimal cluster topology. TiCDC supports multiple downstream platforms, such as TiDB, MySQL, and MQ.
Replicate incremental data using TiDB BinlogDeploy the TiDB Binlog topologySimple TiDB Binlog configuration template (MySQL as downstream)
Simple TiDB Binlog configuration template (Files as downstream)
Full TiDB Binlog configuration template
This is to deploy TiDB Binlog along with the minimal cluster topology.
Use OLAP on SparkDeploy the TiSpark topologySimple TiSpark configuration template
Full TiSpark configuration template
This is to deploy TiSpark along with the minimal cluster topology. TiSpark is a component built for running Apache Spark on top of TiDB/TiKV to answer the OLAP queries. Currently, TiUP cluster's support for TiSpark is still experimental.
Deploy multiple instances on a single machineDeploy a hybrid topologySimple configuration template for hybrid deployment
Full configuration template for hybrid deployment
The deployment topologies also apply when you need to add extra configurations for the directory, port, resource ratio, and label.
Deploy TiDB clusters across data centersDeploy a geo-distributed deployment topologyConfiguration template for geo-distributed deploymentThis topology takes the typical architecture of three data centers in two cities as an example. It introduces the geo-distributed deployment architecture and the key configuration that requires attention.

For more configuration description, see the following configuration examples:

Step 4. Run the deployment command

Before you run the deploy command, use the check and check --apply commands to detect and automatically repair potential risks in the cluster:

  1. Check for potential risks:

    tiup cluster check ./topology.yaml --user root [-p] [-i /home/root/.ssh/gcp_rsa]
  2. Enable automatic repair:

    tiup cluster check ./topology.yaml --apply --user root [-p] [-i /home/root/.ssh/gcp_rsa]
  3. Deploy a TiDB cluster:

    tiup cluster deploy tidb-test v6.3.0 ./topology.yaml --user root [-p] [-i /home/root/.ssh/gcp_rsa]

In the tiup cluster deploy command above:

  • tidb-test is the name of the TiDB cluster to be deployed.
  • v6.3.0 is the version of the TiDB cluster to be deployed. You can see the latest supported versions by running tiup list tidb.
  • topology.yaml is the initialization configuration file.
  • --user root indicates logging into the target machine as the root user to complete the cluster deployment. The root user is expected to have ssh and sudo privileges to the target machine. Alternatively, you can use other users with ssh and sudo privileges to complete the deployment.
  • [-i] and [-p] are optional. If you have configured login to the target machine without password, these parameters are not required. If not, choose one of the two parameters. [-i] is the private key of the root user (or other users specified by --user) that has access to the target machine. [-p] is used to input the user password interactively.

At the end of the output log, you will see Deployed cluster `tidb-test` successfully. This indicates that the deployment is successful.

Step 5. Check the clusters managed by TiUP

tiup cluster list

TiUP supports managing multiple TiDB clusters. The preceding command outputs information of all the clusters currently managed by TiUP, including the cluster name, deployment user, version, and secret key information:

Step 6. Check the status of the deployed TiDB cluster

For example, run the following command to check the status of the tidb-test cluster:

tiup cluster display tidb-test

Expected output includes the instance ID, role, host, listening port, and status (because the cluster is not started yet, so the status is Down/inactive), and directory information.

Step 7. Start a TiDB cluster

Since TiUP cluster v1.9.0, safe start is introduced as a new start method. Starting a database using this method improves the security of the database. It is recommended that you use this method.

After safe start, TiUP automatically generates a password for the TiDB root user and returns the password in the command-line interface.

Method 1: Safe start

tiup cluster start tidb-test --init

If the output is as follows, the start is successful:

Started cluster `tidb-test` successfully. The root password of TiDB database has been changed. The new password is: 'y_+3Hwp=*AWz8971s6'. Copy and record it to somewhere safe, it is only displayed once, and will not be stored. The generated password can NOT be got again in future.

Method 2: Standard start

tiup cluster start tidb-test

If the output log includes Started cluster `tidb-test` successfully, the start is successful. After standard start, you can log in to a database using a root user without a password.

Step 8. Verify the running status of the TiDB cluster

tiup cluster display tidb-test

If the output log shows Up status, the cluster is running properly.

See also

If you have deployed TiFlash along with the TiDB cluster, see the following documents:

If you have deployed TiCDC along with the TiDB cluster, see the following documents: