Nuno Santos created OAK-9930:
--------------------------------
Summary: Thread leak in elastic index statistics module
Key: OAK-9930
URL: https://issues.apache.org/jira/browse/OAK-9930
Project: Jackrabbit Oak
Issue Type: Improvement
Components: indexing
Reporter: Nuno Santos
The method {{ElasticIndexStatistics.CountCacheLoader#reload}} creates a
single-threaded executor to perform a query to Elastic in the background.
However, it does not properly dispose of the thread pool. Notice that the code
does not keep any reference to the executor that it creates, therefore it will
never call shutdown to properly dispose of the executor and of the underlying
thread. This will lead to an accumulation of unused threads, taking up memory
and system resources.
{code:java}
public ListenableFuture<Integer> reload(@NotNull StatsRequestDescriptor crd,
@NotNull Integer oldValue) {
ListenableFutureTask<Integer> task = ListenableFutureTask.create(() ->
count(crd));
Executors.newSingleThreadExecutor().execute(task);
return task;
} {code}
Link to code in Git:
[https://github.com/apache/jackrabbit-oak/blob/5b1916dfd69e82759d80aff867d34bad94ea[…]ackrabbit/oak/plugins/index/elastic/ElasticIndexStatistics.java|https://github.com/apache/jackrabbit-oak/blob/5b1916dfd69e82759d80aff867d34bad94eac760/oak-search-elastic/src/main/java/org/apache/jackrabbit/oak/plugins/index/elastic/ElasticIndexStatistics.java#L192-L195]
In addition to the resource leak, it is not good practice to create a new
thread pool to execute a single request. A better alternative is to either use
the ForkJoin pool provided by the Java runtime or to create a long-lived thread
pool to execute these requests.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
