You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@druid.apache.org by GitBox <gi...@apache.org> on 2018/07/19 19:14:40 UTC

[GitHub] gianm commented on issue #3956: Thread safe reads for aggregators in IncrementalIndex

gianm commented on issue #3956: Thread safe reads for aggregators in IncrementalIndex
URL: https://github.com/apache/incubator-druid/pull/3956#issuecomment-406384627
 
 
   I was just looking at this issue again after the conversations on the mailing list about sketch synchronization: https://lists.apache.org/thread.html/9899aa790a7eb561ab66f47b35c8f66ffe695432719251351339521a@%3Cdev.druid.apache.org%3E
   
   I was wondering, does it make more sense for thread-safety here to be handled systematically (at the IncrementalIndex) or for each aggregator to be thread safe? Currently we do different approaches: the sketch aggregators endeavor to be thread-safe on their own. The primitive aggregators don't bother to even try, and they're probably fine, since they're primitives. HyperLogLogAggregator tries a little bit -- it at least makes sure the different calls use different buffer objects -- but I bet it has a bug where "get" could potentially read something weird and corrupt in some rare situations. (Like if the offset is being updated while a "get" is going on.)

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

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