MartijnVisser opened a new pull request, #29280:
URL: https://github.com/apache/flink/pull/29280

   ## What is the purpose of the change
   
   * Prevent `compile_cron_azure` on release-1.20 from running out of heap in 
japicmp on `flink-s3-fs-presto`
   
   It failed that way on 80 of the 102 nightlies that ran since 2026-06-07, the 
oldest build Azure still has. The leg compiles on JDK 8 on a Microsoft-hosted 
agent, where Maven gets a 1992 MiB heap. About 850 MiB of that is already live 
when the build reaches presto, and on its own the comparison of the 97 MB 
shaded jar fails at a 1000 MiB heap. master, release-2.3 and release-2.2 run 
this leg on JDK 17 and pass.
   
   The only `@PublicEvolving` classes in that jar, `S3RecoverableWriter` and 
`S3RecoverableFsDataOutputStream`, come from `flink-s3-fs-base`, which 
`flink-s3-fs-hadoop` also bundles and checks.
   
   ## Brief change log
   
   * Set `japicmp.skip` in `flink-s3-fs-presto`
   
   ## Verifying this change
   
   Replayed the `compile_cron_azure` Maven call on JDK 8 with 
`-XX:MaxRAM=8128880k -XX:ActiveProcessorCount=2`, which gives the agent's 1992 
MiB heap:
   
   * release-1.20 (16b92f3d6a3): 2 of 3 runs fail with `OutOfMemoryError: Java 
heap space` in japicmp on presto
   * this branch: 5 of 5 runs pass all 179 modules, and so do runs with 
`-Xmx1850m` and `-Xmx1750m`
   
   Two of those five runs still reached the heap limit for a few seconds in 
`flink-python`'s japicmp and recovered, so the leg stays close to its heap.
   
   Making `S3RecoverableWriter` final still fails the build, in 
`flink-s3-fs-hadoop` (`CLASS_NOW_FINAL`). On 16b92f3d6a3, presto's japicmp 
reports the same break and nothing else. The presto jar is unchanged.
   
   ## Does this pull request potentially affect one of the following parts:
   
     - Dependencies (does it add or upgrade a dependency): no
     - The public API, i.e., is any changed class annotated with 
`@Public(Evolving)`: no
     - The serializers: no
     - The runtime per-record code paths (performance sensitive): no
     - Anything that affects deployment or recovery: JobManager (and its 
components), Checkpointing, Kubernetes/Yarn, ZooKeeper: no
     - The S3 file system connector: no
   
   ## Documentation
   
     - Does this pull request introduce a new feature? no
     - If yes, how is the feature documented? not applicable
   
   ---
   
   ##### Was generative AI tooling used to co-author this PR?
   
   - [X] Yes (please specify the tool below)
   
   Generated-by: Claude Code (Claude Opus 5.5)
   


-- 
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