Skip to content

Allow external access to the HDFS cluster #174

Closed
@nightkr

Description

@nightkr

Currently we just expose overlay network addresses in both hdfs-site.xml and dynamic registrations to the namenodes. This prevents users from accessing HDFS from outside of the k8s overlay network.

This requires:

  • Namenodes must be accessed using stable addresses (such as a LoadBalancer Service where available)
    • The generated discovery profile must be altered to use this address
  • Datanodes must be accessed using their associated NodePort Service addresses

Datanode addresses cannot currently be overridden in HDFS, I have outlined the changes required for this in https://github.com/stackabletech/hadoop/tree/spike/override-datanode-id. This is a purely server-side change.

I have a spike that implements this semi-manually in https://github.com/stackabletech/hdfs-operator/tree/spike/external-access, if we consider that access pattern acceptable then we can start looking into how to integrate it better into both HDFS and the operator.

Metadata

Metadata

Assignees

Labels

customer-requestrelease-noteDenotes a PR that will be considered when it comes time to generate release notes.release-note/action-requiredDenotes a PR that introduces potentially breaking changes that require user action.release/24.3.0

Type

No type

Projects

Status

Done

Status

Done

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions