michaelschiff commented on PR #14916: URL: https://github.com/apache/druid/pull/14916#issuecomment-1700058735
> I think there are many places in the code which assume that when we are using hashed or range, forceGuaranteedRollup must be true. Breaking that assumption might cause erratic behaviour in some cases. I wasn't aware of this assumption elsewhere, but if thats the case then I think those places should probably change - so I'd go with your option 1 It seems reasonable to me that there are use-cases where its desirable to retain un-aggregated data in druid, but where you would still want to take advantage of the performance improvements that come from hash/range based partitioning. The "no rollup" use-case is also described in the docs https://druid.apache.org/docs/latest/ingestion/rollup/#:~:text=When%20you%20disable%20rollup%2C%20Druid,is%2C%20without%20any%20rollup%20summarization and it seems weird that you can disable rollup, but that if you then want to re-organize your partitioning you're not able -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
