[jira] [Updated] (HBASE-27943) Rework kuttl image as a yetus precommit base

["Nick Dimiduk (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HBASE-27943?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Nick Dimiduk updated HBASE-27943:
---------------------------------
    Status: Patch Available  (was: In Progress)

> Rework kuttl image as a yetus precommit base
> --------------------------------------------
>
>                 Key: HBASE-27943
>                 URL: https://issues.apache.org/jira/browse/HBASE-27943
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Nick Dimiduk
>            Assignee: Nick Dimiduk
>            Priority: Major
>
> Due to permissions issues on build workers (INFRA-24621), I think we'll benefit by shifting the perspective a little on how the kuttl image is used in CI. The current PRs establish the kuttl image as a (relatively) small, self-contained utility image. When it is invoked in CI, we run yetus with docker-in-docker support, and launch tests by calling `docker container run ... kuttl ...`. So, docker is invoked from within the Yetus pre-commit docker container. Rather, I want to implement the kuttl image as extending from the yetus image (or yetus-base, I'm not sure yet). That way we don't need to run pre-commit with docker-in-docker mode and the precommit process can invoke kuttl directly.
> The existing kuttl image build is too sophisticated to be invoked by Yetus as a provided docker image (it makes use of build-args). It may also be too sophisticated to be run in a Jenkins worker (it makes use of buildx, INFRA-24704). So I guess we need to publish the image someplace and use the published tag via the Yetus `--docker-tag` flag.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)