These images are part of the Bigdata docker image series. All of the images use the same base docker image which contains plugin scripts to launch different project in containerized environments.
For more detailed instruction about the available environment variables see the README in the flokkr/docker-baseimage repository.
Docker images are tested with Kubernetes
The easiest way to start is to do a kubectl apply -f . from the ./exmaples directories (Using ephemeral storage!)
For more specific use case it's recommended to use flekszible. The resource definitions can be found in this repository (./hadoop,./hdfs,./yarn...)
Install Flekszible (download binary and put it to the path)
- Create a working dir
cd /tmp
mkdir cluster
cd cluster
- Add this repository as a source
flekszible source add github.com/flokkr/docker-hadoop
- Choose and add required services:
flekszible app add hdfs
- Generate Kubernetes resource files
flekszible generate
- Lunch the rockets:
kubectl apply -f .
You can list available apps (after source import):
flekszible app search
+---------+-------------------------------+
| path | description |
+---------+-------------------------------+
| hdfs | Apache Hadoop HDFS base setup |
| hdfs-ha | Apache Hadoop HDFS, HA setup |
...
The base setup can be modified with additional transformatios:
flekszible definitions search | grep hdfs
...
| hdfs/persistence | Add real PVC based persistence |
| hdfs/onenode | remove scheduling rules to make it possible to run multiple datanode on the same k8s node. |
...
You can apply transformations with modifing the Flekszible descriptor file:
Original version:
source:
- url: github.com/flokkr/docker-hadoop
import:
- path: hdfsModified:
source:
- url: github.com/flokkr/docker-hadoop
import:
- path: hdfs
transformations:
- type: hdfs/onenode
- type: image
image: flokkr/hadoop:3.2.0