// the find
wgzhao/Addax
Actively maintained successor to Alibaba DataX — a fast, versatile, open-source ETL tool for 20+ RDBMS and NoSQL data sources.
Addax is a Java batch ETL tool forked from Alibaba's DataX, which has been frozen since 2023. You describe a copy as a JSON job with one reader and one writer, and it moves records between them using plugins for relational, NoSQL, Hadoop-side, and file sources. It suits teams with existing DataX jobs or a need for scripted batch copies between databases, HDFS, and Hive without building a pipeline framework.
The job format keeps DataX's reader/writer shape, and the example jobs under core/src/main/job show it clearly, so existing configs are a reasonable starting point. The e2e directory is the part I'd point people to: each case pairs a job.json with setup and verify SQL or shell scripts, covering MySQL and PostgreSQL round trips, HDFS text, Parquet and ORC output, and datetime and timezone handling, and a workflow runs them. That's where type-conversion regressions get caught. The release process is written down in RELEASING.md, and a commit-lint workflow sits alongside the Maven publish CI, which makes the maintenance claims easier to check.
Reference documentation lives in a separate repo (addax-docs), so reading the options for a given plugin at the version you have pinned means cloning a second project. The coverage numbers don't agree: the repo description says 20+ sources, the README says 30+, and the logo grid shows 34, several of which are file formats such as dBase, Excel, and Access rather than servers. The maintenance story is a promise in the README: monthly releases and issue triage 'within days' are claims the releases page and issue tracker have to back up. New sources are only added on demonstrated community demand, so an unlisted source means writing the plugin yourself.