In FlinkSQL, when there is data skew issues with joins (i.e. not a normal distribution of keys), is there a way to force different data shuffling/distribution? We have a few joins that unfortunately are mostly performed on a single worker and wondering what are some possible solutions. Is there a way to force a better distribution of both sides of the join?

Perhaps one of the join hints can help in your case, see https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sql/queries/hints/#join-hints

Unfortunately those are all for batch mode