[GitHub] [druid] loquisgon commented on pull request #12367: Do not select intervals with only tombstones for compaction in auto-compaction

[GitBox <gi...@apache.org>]

loquisgon commented on PR #12367:
URL: https://github.com/apache/druid/pull/12367#issuecomment-1087827009

   > The real bug here is that the job is failing when given all tombstones. Skipping the tombstones as defined here is going to create other oddities, like, if a tombstone partially overlaps another segment, the compaction job for that chunk of time won't include the tombstone to overlap. So, now that job hopefully will be given only a partial interval, but when it's given a partial interval, that creates a new chance for corner cases where it might or might not actually produce the correct output segment because it wasn't actually given all of the segments.
   > 
   > Tombstone segments are segments just like any other segment, they just happen to have no data in them. They should not be special-cased anywhere, including when the coordinator is picking segments to compact.
   
   Replacing this approach with https://github.com/apache/druid/pull/12392 which actually considers and do not skip tombstones for Druid datasource ingestion. Closing this one.


-- 
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: commits-unsubscribe@druid.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org