Documentation Index
Fetch the complete documentation index at: https://private-7c7dfe99-page-updates.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
Install ClickHouse on rpm-based distributions
It is recommended to use official pre-compiled rpm packages for CentOS, RedHat, and all other rpm-based
Linux distributions.
Review recommendations
Before installing ClickHouse, review the following recommendations:- Swap: Disable the operating system’s swap file in production environments.
- Disk space: The ClickHouse binary requires at least 2.5 GB of disk space for installation.
- Network: For distributed deployments (clustering), use at least 10 Gbit network connectivity. Network bandwidth is critical for processing distributed queries with large amounts of intermediate data, as well as for replication.
- Estimate data volume: Take a sample of your data and calculate the average row size, then multiply by the number of rows you plan to store.
- Apply the compression coefficient: Load a sample into ClickHouse and compare the original data size with the stored table size. Clickstream data, for example, is typically compressed 6-10x.
- Account for replicas: If you plan to store data in multiple replicas, multiply the estimated volume by the number of replicas.
Set up the RPM repository
Add the official repository by running the following command:zypper package manager (openSUSE, SLES), run:yum install can be replaced by zypper install, depending
on which package manager you’re using.Install ClickHouse server and client
To install ClickHouse run the following commands:- You can replace
stablewithltsto use different release kinds based on your needs. - You can download and install packages manually from packages.clickhouse.com/rpm.
- To specify a particular version, add
-$versionto the end of the package name, for example:
Start ClickHouse server
To start ClickHouse server, run:Install standalone ClickHouse Keeper
To installclickhouse-keeper on standalone ClickHouse Keeper servers, run: