mjsax commented on a change in pull request #10917:
URL: https://github.com/apache/kafka/pull/10917#discussion_r660334734
##########
File path:
streams/src/main/java/org/apache/kafka/streams/kstream/internals/KStreamKStreamJoin.java
##########
@@ -186,12 +190,17 @@ public void process(final K key, final V1 value) {
@SuppressWarnings("unchecked")
private void emitNonJoinedOuterRecords(final
WindowStore<KeyAndJoinSide<K>, LeftOrRightValue> store) {
+ if (minTime.minTime >= maxObservedStreamTime.get() - joinAfterMs -
joinBeforeMs - joinGraceMs) {
Review comment:
It's a performance question. Without `joinBeforeMs` we only get 3K
rec/sec throughput in our benchmarks, while we get 12K rec/sec throughput with
`joinBeforeMs`.
##########
File path:
streams/src/main/java/org/apache/kafka/streams/kstream/internals/KStreamKStreamJoin.java
##########
@@ -186,12 +190,17 @@ public void process(final K key, final V1 value) {
@SuppressWarnings("unchecked")
private void emitNonJoinedOuterRecords(final
WindowStore<KeyAndJoinSide<K>, LeftOrRightValue> store) {
+ if (minTime.minTime >= maxObservedStreamTime.get() - joinAfterMs -
joinBeforeMs - joinGraceMs) {
+ return;
+ }
+
try (final KeyValueIterator<Windowed<KeyAndJoinSide<K>>,
LeftOrRightValue> it = store.all()) {
while (it.hasNext()) {
final KeyValue<Windowed<KeyAndJoinSide<K>>,
LeftOrRightValue> record = it.next();
final Windowed<KeyAndJoinSide<K>> windowedKey = record.key;
final LeftOrRightValue value = record.value;
+ minTime.minTime = windowedKey.window().start();
Review comment:
Well, updating `minTime` does not cost us anything because it' a plain
in-memory operation (also no additional serde costs).
--
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]
