wuchong opened a new issue, #2683:
URL: https://github.com/apache/fluss/issues/2683

   ### Search before asking
   
   - [x] I searched in the [issues](https://github.com/apache/fluss/issues) and 
found nothing similar.
   
   
   ### Motivation
   
   The `fluss-quickstart-flink` Docker image was previously provided as a 
convenient, self-contained way for new users to quickly try out Fluss with 
Apache Flink through the quickstart documentation. However, this image is no 
longer shipped or referenced in the current quickstart guide (see 
https://github.com/apache/fluss/pull/1997).
   
   While testing Fluss 0.9 RC1 with the updated quickstart instructions, I 
encountered significant usability issues during the JAR artifact download 
phase. Due to an unstable network connection (very common in mainland China), 
several `curl` commands failed silently—terminating prematurely without clear 
error messages but leaving partially downloaded (and thus corrupted) JAR files 
on disk. Believing the setup was complete, I proceeded to run `docker-compose`, 
only to encounter cryptic “class not found” errors. After extensive debugging, 
I discovered that some JARs were either missing or incomplete. 
   
   This experience highlights a critical flaw: while the current 
approach—downloading JARs separately and mounting them into a generic Flink 
Docker image—offers flexibility in choosing Flink versions, it severely 
compromises the core purpose of a *quickstart*: to enable anyone, regardless of 
prior Flink or Fluss experience, to get up and running within five minutes with 
zero friction.
   
   In contrast, the Apache Iceberg project provides a prebuilt `spark-iceberg` 
Docker image (https://github.com/databricks/docker-spark-iceberg), which 
bundles all necessary dependencies and dramatically lowers the barrier to entry 
(see https://iceberg.apache.org/spark-quickstart/#docker-compose). This pattern 
demonstrates that a dedicated quickstart image is not just helpful—it’s 
essential for a smooth onboarding experience.
   
   Given that Fluss 0.9 now includes Spark connector support, we should also 
consider publishing a `fluss-quickstart-spark` image in the future.
   
   
   
   ### Solution
   
   That said, the current JAR-download + flexible-Flink-image approach still 
has its place—for advanced users who need version control or customization. 
However, it belongs in a more detailed “Engine Flink → Getting Started” section 
(which currently uses standalone deployment). We could enhance that section by 
adding a tabbed option for Docker Compose–based deployment. 
   
   In summary, I propose:
   1. Reintroducing a prebuilt `fluss-quickstart-flink` Docker image (ideally 
based on Flink 2.2 to showcase the latest features) for the main quickstart.
   2. Moving the current manual JAR + Docker Compose workflow to an advanced 
“Engine Flink → Getting Started” guide.
   3. Planning a `fluss-quickstart-spark` image alongside for Spark Quickstart. 
   
   Note: this is not a blocker of release 0.9, and should be done after 0.9 is 
released. 
   
   ### Anything else?
   
   _No response_
   
   ### Willingness to contribute
   
   - [ ] I'm willing to submit a PR!


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to