atris commented on a change in pull request #815: LUCENE-8213: Introduce
Asynchronous Caching in LRUQueryCache
URL: https://github.com/apache/lucene-solr/pull/815#discussion_r324495925
##########
File path: lucene/core/src/java/org/apache/lucene/search/LRUQueryCache.java
##########
@@ -726,14 +794,35 @@ public ScorerSupplier scorerSupplier(LeafReaderContext
context) throws IOExcepti
DocIdSet docIdSet;
try {
docIdSet = get(in.getQuery(), cacheHelper);
+ } catch (AsyncQueryLoadInProgressException e) {
+ // Query is being loaded asynchronously. Use uncached version but
+ // do not do a new load of the same query into the cache
+ return in.scorerSupplier(context);
} finally {
lock.unlock();
}
if (docIdSet == null) {
if (policy.shouldCache(in.getQuery())) {
- docIdSet = cache(context);
- putIfAbsent(in.getQuery(), docIdSet, cacheHelper);
+ // If asynchronous caching is requested, perform the same and return
+ // the uncached iterator
+ if (executor != null) {
+ FutureTask<Void> task = new FutureTask<>(() -> {
+ DocIdSet localDocIdSet = cache(context);
+ putIfAbsent(in.getQuery(), localDocIdSet, cacheHelper);
+
+ //remove the key from inflight -- the key is loaded now
+ inFlightAsyncLoadQueries.remove(in.getQuery());
+ return null;
+ });
+ inFlightAsyncLoadQueries.add(in.getQuery());
Review comment:
Yes, that is precisely the idea. The alternative to that is a more
aggressive locking which can potentially have a larger negative effect than a
thread's caching effort. Thanks for raising this, I have added docs to explain
this now.
----------------------------------------------------------------
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:
us...@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
