[HIVE-28324] HIVE_CLUSTER_ID in env: unified way to mark a cluster

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.1.0
• Labels: pull-request-available
• Assignee: László Bodor
• Reporter: László Bodor
• Created: 12/Jun/24 14:28
• Updated: 20/Jun/24 07:54
• Resolved: 17/Jun/24 08:05

Description

The proposal is to introduce an environment variable that can be injected by the environment to hive components to be able to identify a cluster. There are very different hive setups all over the world and all of them have to implement such things sooner or later. In order to make the hive vendors' life easier, let's support a common name for this: cluster id. This doesn't prevent anyone from using their own.
What hive can take care of is to propagate this value to different places in the code where it makes sense (hs2, llap daemon)

Following naming conventions, this can be interpreted as:
environment variable: HIVE_CLUSTER_ID
command line java property: hive.cluster.id (as suggested here)

Precedence order is: cli prop, env var

Issue Links

• is related to: HIVE-28095 Hive Query History
• links to: GitHub Pull Request #5304