Closed yanghua closed 10 months ago
@pan3793 WDYT?
I write my original thoughts in README:
... we may want to leverage the template engine like Ansible plus Jinja2 to generate the docker-compose.yml file and configuration files for each component, so that user could easily customize the testing environment by modifying the configurations.
In fact, a lot of Hadoop ecosystem components usually mix-deploys in a group of machines, we can not achieve that by using docker-compose profile
I write my original thoughts in README:
... we may want to leverage the template engine like Ansible plus Jinja2 to generate the docker-compose.yml file and configuration files for each component, so that user could easily customize the testing environment by modifying the configurations.
In fact, a lot of Hadoop ecosystem components usually mix-deploys in a group of machines, we can not achieve that by using docker-compose profile
Agree, the mechanism of profiles can only solve some issues. Its flexibility is limited. My original thought only is for picking some of those components that we have supported for different scenarios.
As adding more components to Hadoop testing, it would be better to introduce a solution to isolate and compose different component groups, e.g (mini-hadoop-cluster, lakehouse-iceberg, and so on).
IMO, the docker-compose profile feature is a good choice, just like the maven profile.