I'm testing the new feature for faster checkpoint ...
# troubleshooting
j
I'm testing the new feature for faster checkpoint recovery using s5cmd on Flink 2.0 but I get an error when my jobs are initializing and recovering their data:
s3 url should start with "s3://"
Here's the relevant configurations (I left out some configs):
Copy code
s3:
  s5cmd:
  path: /opt/flink/bin/s5cmd
execution:
  checkpointing:
    dir: <s3p://flink-p/checkpoints>
I use s3p filesystem for checkpointing and s3a for sinking to parquet files on s3 but it seems pretty clear that s5cmd needs a proper s3:// url