For extra context, I'm observing checkpoint failures like this one:

```
java.io.IOException: Could not flush to file and close the file system
output stream to s3://xyz-flink/my-job/checkpoints/*** in order to obtain
the stream state handle
```

on multiple jobs. I wonder whether switching over from pesto to s3a could
help, but the docs:

https://nightlies.apache.org/flink/flink-docs-release-1.20/docs/deployment/filesystems/s3/#hadooppresto-s3-file-systems-plugins

recommend the presto S3 filesystem for checkpointing to S3 so I guess no.

Another setting that I'm trying is:

```
execution.checkpointing.file-merging.enabled
```

Will share my findings—if any.

Reply via email to