+1 Thank you Samrat for the proposal, and Piotr, Aleksandr and Gabor for the detailed feedback.
>From my perspective most of the key details of the FLIP have already been addressed, and would be happy if we could proceed with Phase 1 (adding the native s3 connector as optional plugin for community validation) as suggested by Samrat soon. Thanks, Marton On Mon, Feb 16, 2026 at 7:00 PM Aleksandr Iushmanov <[email protected]> wrote: > Hi Samrat, > > Thank you for incorporating the suggestions and answering my questions! > > I have follow up proposal on below: > > > To execute the migration for any existing Job, an engineer only needs to > > perform a "JAR-swap" in the Flink plugins/ directory: > > > This would require potentially error prones config switchovers and > replacement of the flink image in production. I would prefer to have a > feature flag that would allow flink to choose which of 2 plugins to load if > both are provided. I have put up a proposal PR [1] that introduces a config > allowing to package both jars in the image and put migration under a simple > feature flag without the need to worry about schema compatibility and this > approach would work for all clouds. The only thing that would be needed for > this approach to work smoothly is to override the `getPriority` method in > the newly introduced FS to `return -1`. WDYT? > > [1] https://github.com/apache/flink/pull/27617 > -- > Kind regards, > Alex >
