By default, HDFS is not used or installed with Metering.
However, if you set the spec.unsupportedFeatures.enableHDFS
unsupported features toggle to true, then you can install a non-production grade HDFS cluster alongside metering for storage.
We are not HDFS experts, and thus we will not directly support HDFS, and it should only be used for testing or development. Currently there is no way to secure the communications with HFDS, meaning all communications are done in plain text, without authentication.
Proceed at your own risk.
hdfs-namenode-data-hdfs-namenode-0
Used by the hdfs-namenode pod to store metadata about the files and blocks stored in the hdfs-datanodes.- One
hdfs-datanode-data-hdfs-datanode-$i
PV per hdfs-datanode replica. Used by each hdfs-datanode pod to store blocks for files in the HDFS cluster.
Use hdfs-storage.yaml as a template and adjust the class: null
value to name of the StorageClass
to use.
Leaving the value null
will cause Metering to use the default StorageClass for the cluster.
spec.hadoop.spec.hdfs.datanode.storage.class
spec.hadoop.spec.hdfs.namenode.storage.class
Use hdfs-storage.yaml as a template and adjust the size: "5Gi"
value to the desired capacity for the following sections:
spec.hadoop.spec.hdfs.datanode.storage.size
spec.hadoop.spec.hdfs.namenode.storage.size