Question about Flink checkpointing: If Flink checkpoints while streaming data from S3, does it: 1. Save the file’s state (filenames, timestamp etc.) in checkpoints? 2. Can this state be read manually to extract some of the checkpoint metadata if needed?

Checkpoints are for Flink internal usage. You might be able to inspect them using the State Processor API but there are no guarantees around that

Regarding the FileSource, it does persist file paths in checkpoints / savepoints.

Thanks, is there any documentation what other things are stored while checkpointing? couldn’t find it here

No, that’s not documented

@sap1ens do you know if we can ready the file paths in checkpoints through the State Processor API?

Pretty sure it’s possible, but State Processor API is not easy to use, you need to reverse-engineer Flink source code to understand names and types of the state variables.