[HDDS-5114] Do not store datanode details in SCM pipeline db

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Labels: kubernetes
• Assignee: Unassigned
• Reporter: Marton Elek
• Created: 19/Apr/21 09:01
• Updated: 11/Mar/22 06:08

Description

When the IP addresses of a cluster has been changed, the same datanodes will be connected with a new ip address to the SCM. SCM assumes that the nodes are up (which is right) and pipelines are available. But when allocateBlock is served the persisted pipeline information is used instead of the fresh ones.

It should be enough to save the node uuid in the pipeline information and all the other data can be excluded (ip and ports) and use the latest node information.

Issue Links

• is a child of: HDDS-6438 Support Ozone to run as a service within Kubernetes