scwhittle commented on a change in pull request #16495:
URL: https://github.com/apache/beam/pull/16495#discussion_r787112419
##########
File path:
sdks/java/harness/src/main/java/org/apache/beam/fn/harness/Caches.java
##########
@@ -165,10 +141,57 @@ public int weigh(Object key, Object value) {
}
@VisibleForTesting
- static Cache<Object, Object> forCache(
-
org.apache.beam.vendor.guava.v26_0_jre.com.google.common.cache.Cache<CompositeKey,
Object>
- cache) {
- return new SubCache<>(cache, CompositeKeyPrefix.ROOT);
+ static <K, V> Cache<K, V> forMaximumBytes(long maximumBytes) {
+ // We specifically use Guava cache since it allows for recursive
computeIfAbsent calls
+ // preventing deadlock from occurring when a loading function mutates the
underlying cache
+ LongAdder weightInBytes = new LongAdder();
+ return new SubCache<>(
+ new ShrinkOnEviction(
+ CacheBuilder.newBuilder()
+ .maximumWeight(maximumBytes >> WEIGHT_RATIO)
+ .weigher(
+ new Weigher<Object, WeightedValue<Object>>() {
+
+ @Override
+ public int weigh(Object key, WeightedValue<Object>
value) {
+ // Round up to the next closest multiple of
WEIGHT_RATIO
+ long size = ((value.getWeight() - 1) >>
WEIGHT_RATIO) + 1;
Review comment:
It seems like we used to have the cache weigh the key with
Caches.weigh(key) but I no longer see that happening. Was that intentional?
If not, add a test where there are large keys to verify that they are
weighed.
--
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]
