gtwuser commented on issue #8532: URL: https://github.com/apache/hudi/issues/8532#issuecomment-1562722125
@ad1happy2go job completed in 3hrs, on doing the following: 1. "hoodie.bulkinsert.shuffle.parallelism" updated to `5` instead of `200` . 2. reducing 20 DPUs to 5 DPUs for glue job 3. removing these configs `--conf spark.yarn.executor.memory=8g --conf spark.yarn.executor.memory=10g --conf spark.driver.memory=10g` 4. Avoiding `dataframe.count()` calls 5. hoodie.parquet.max.file.size set to `12MB` But im curious to understand and was even looking for `detailed` documentation on 1. how is `hoodie.parquet.max.file.size` and `shuffle.parallelism` related ? 2. what causes `shuffling` of data 3. what component or parameter we are actually playing with while configuring `hoodie.bulkinsert.shuffle.parallelism`. 4. And finally one simple example on how setting it to huge number would lower the performance. ^^ @ad1happy2go @nsivabalan @umehrot2 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
