pphust commented on a change in pull request #8323: Make sure
ReferenceCountingSegment.decrement() is invoked correctly when useCache=true
URL: https://github.com/apache/incubator-druid/pull/8323#discussion_r315527431
##########
File path:
server/src/main/java/org/apache/druid/segment/realtime/appenderator/SinkQuerySegmentWalker.java
##########
@@ -231,7 +231,7 @@ public SegmentDescriptor apply(final PartitionChunk<Sink>
chunk)
// 1) Only use caching
if data is immutable
// 2) Hydrants are not
the same between replicas, make sure cache is local
if
(hydrantDefinitelySwapped && cache.isLocal()) {
- QueryRunner<T>
cachingRunner = new CachingQueryRunner<>(
+ QueryRunner<T>
cachingRunner = new CloseableCachingQueryRunner<>(
Review comment:
It looks a good idea and I have modified the code correspondingly.
I am not very sure about "Reduce the level of nesting". I think at least we
need 2-level cycles in code. One for each sink/segment in time line and the
other for each hydrant in sink.
I just change `FunctionalIterable.create(specs).transform()` to
`Iterables.transform()` currently. Is it what you want?
Or I need to separate the second cycle, here I mean code after `return new
SpecificSegmentQueryRunner<>` , into one different method like
`getQueryRunnerForHydrants(Sink)` and it would make code look clear.
FYI, the new resolution has been cherry-picked into 0.15.1-incubating branch
locally and it works well at our cluster.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
