damccorm opened a new pull request, #41071: URL: https://github.com/apache/beam/pull/41071
## Summary Fixes transient `java.net.SocketTimeoutException: Connect timed out` failures in `PostCommit Java SingleStoreIO IT` (e.g., https://github.com/apache/beam/actions/runs/42364200637 and https://github.com/apache/beam/actions/runs/42203139180). ### Root Cause In `.github/workflows/beam_PostCommit_Java_SingleStoreIO_IT.yml`, `kubectl wait svc/svc-sdb-cluster-ddl --for=jsonpath='{.status.loadBalancer.ingress[0].ip}'` completes as soon as GKE allocates an external LoadBalancer IP, before the GCP LoadBalancer forwarding rule and `kube-proxy` health checks finish programming routes to the SingleStore pod on TCP port `3306`. When the Gradle integration test starts quickly (e.g. on a Gradle cache hit), `SingleStoreIOConnectionAttributesIT` times out attempting to open a TCP socket to `<lb_ip>:3306`. ### Fix Wait up to 5 minutes (`60 * 5s`) for `$loadbalancer_IP:3306` to accept TCP connections via `nc -zv -w 5 "$loadbalancer_IP" 3306` in the `Install SingleStore cluster` workflow step before launching the Gradle integration test. ## Verification - Triggered `PostCommit Java SingleStoreIO IT` on this branch: https://github.com/apache/beam/actions/runs/42366624917 -- 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]
