bbb4aaa opened a new pull request, #2290:
URL: https://github.com/apache/fluss/pull/2290
### Purpose
The purpose of this change is to fix a connectivity issue between clients
and the coordinator-server when Fluss is deployed via Helm. Previously, the
`INTERNAL` listener was bound to `0.0.0.0`, which caused the server to register
itself with an ambiguous address. This prevented clients from correctly
resolving the coordinator-server location during the `CREATE CATALOG` process
via `bootstrap.servers`.
### Brief change log
- Modified `helm/templates/sts-coordinator.yaml` to change the `INTERNAL`
listener bind address from `0.0.0.0` to `${POD_IP}`.
- This ensures that the coordinator-server registers its specific Pod IP for
internal cluster communication, allowing metadata requests and bootstrap
connections (e.g., `coordinator-server-hs.fluss:9123`) to succeed.
### Tests
- Verified by deploying Fluss in a Kubernetes cluster using the updated Helm
chart.
- Confirmed that Flink SQL `CREATE CATALOG` commands now succeed without
"connection refused" or "metadata not available" errors.
- Manual verification of `$FLUSS_HOME/conf/server.yaml` inside the
coordinator pod to ensure the `bind.listeners` property is correctly rendered
with the Pod's IP.
### API and Format
- This change does not affect public APIs or storage formats. It only
modifies the network binding configuration in the Kubernetes deployment
template.
### Documentation
- No changes to formal documentation are required as this is a bug fix for
the deployment configuration.
--
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]
