4717987e37
### Why are the changes needed? This change addresses an issue where the Flink engine in Kyuubi would perform a Hadoop classpath check even when a `flink-shaded-hadoop` jar is already present in the Flink `lib` directory. In such cases, the check is unnecessary and may cause confusion or warnings in environments where the shaded jar is used instead of a full Hadoop classpath. By skipping the check when a `flink-shaded-hadoop` jar exists, we improve compatibility and reduce unnecessary log output. ### How was this patch tested? The patch was tested by deploying Kyuubi with a Flink environment that includes a `flink-shaded-hadoop` jar in the `lib` directory and verifying that the classpath check is correctly skipped. Additional tests ensured that the check still occurs when neither the Hadoop classpath nor the shaded jar is present. Unit tests and manual verification steps were performed to confirm the fix. ### Was this patch authored or co-authored using generative AI tooling? No. Closes #7113 from cutiechi/fix/flink-classpath-missing-hadoop-check. Closes #7113 99a4bf834 [cutiechi] fix(flink): fix process builder suite 7b9998760 [cutiechi] fix(flink): remove hadoop cp add ea33258a3 [cutiechi] fix(flink): update flink hadoop classpath doc 6bb3b1dfa [cutiechi] fix(flink): optimize hadoop class path messages c548ed6a1 [cutiechi] fix(flink): simplify classpath detection by merging hasHadoopJar conditions 9c16d5436 [cutiechi] Update kyuubi-server/src/main/scala/org/apache/kyuubi/engine/flink/FlinkProcessBuilder.scala 0f729dcf9 [cutiechi] fix(flink): skip hadoop classpath check if flink-shaded-hadoop jar exists Authored-by: cutiechi <superchijinpeng@gmail.com> Signed-off-by: Cheng Pan <chengpan@apache.org> |
||
|---|---|---|
| .. | ||
| spark | ||
| engine_lifecycle.md | ||
| engine_on_kubernetes.md | ||
| engine_on_yarn.md | ||
| engine_share_level.md | ||
| high_availability_guide.md | ||
| hive_metastore.md | ||
| index.rst | ||
| kyuubi_on_kubernetes.md | ||
| migration-guide.md | ||