This section describes how to install CrateDB on a single node. In order to get a CrateDB cluster up and running it is enough to repeat the process on another machine if the network supports multicast. The Nodes will discover and form a cluster automatically. Just like magic.
In case the network doesn’t support multicast, a unicast configuration is required. Please refer to our article on Multi Node Setup in order to learn how to setup the unicast configuration.
We recommend installing Java 8 update 20 or later.
After downloading the latest stable version of CrateDB from https://crate.io/download/ the package needs to be unpacked:
sh$ tar zxf crate-*.tar.gz sh$ cd crate-*
Once unpacked CrateDB can be started in the foreground like this:
In order to start in the background add the
-d option. If started in the
foreground you can use
Control-C to stop the process again.
The CrateDB Shell (Crash) is shipped with the tarball distribution.
You can start the shell using the
The CrateDB distribution also comes with a web based administration
interface. It is served on port
4200 and you can access it with
your browser once Crate is started:
The admin interface runs on each node where CrateDB is installed. If the previous URL is accessed via libraries (eg. cURL), or via offline browsers (eg. wget) the request will be handled by the CrateDB Rest API, and the response will be JSON.
Docker is a lightweigt runtime to run distributed applications in a sandboxed container. We provide a Docker image for CrateDB for easy installation:
sh$ docker pull crate sh$ docker run -d -p 4200:4200 -p 4300:4300 crate crate -Ccluster.name=my_cluster
We provide distributions for RHEL/YUM based systems, Ubuntu, Debian and Arch Linux.
You can find instruction how to install CrateDB for production on the Download page.
In case your distribution is missing feel free to contact us. We’re happy to create and maintain additional packages if we see enough interest.