Hadoop should be entirely built from source, so that patches can be applied. Currently, only the HDFS native client is built to create the fuse-dfs executable.
Instead of building it in the java-base image, it could be easier to use the Dockerfile provided in the Hadoop sources (https://github.com/apache/hadoop/blob/rel/release-3.3.6/dev-support/docker/Dockerfile). One reason is that the libtirpc-devel package is required which is not available in the used repositories.
Hadoop should be entirely built from source, so that patches can be applied. Currently, only the HDFS native client is built to create the fuse-dfs executable.
Instead of building it in the java-base image, it could be easier to use the Dockerfile provided in the Hadoop sources (https://github.com/apache/hadoop/blob/rel/release-3.3.6/dev-support/docker/Dockerfile). One reason is that the libtirpc-devel package is required which is not available in the used repositories.