siddharthteotia commented on code in PR #11558:
URL: https://github.com/apache/pinot/pull/11558#discussion_r1322626318
##########
pinot-segment-local/src/main/java/org/apache/pinot/segment/local/realtime/impl/invertedindex/RealtimeLuceneTextIndex.java:
##########
@@ -109,27 +112,43 @@ public ImmutableRoaringBitmap getDictIds(String
searchQuery) {
@Override
public MutableRoaringBitmap getDocIds(String searchQuery) {
MutableRoaringBitmap docIDs = new MutableRoaringBitmap();
- Collector docIDCollector = new RealtimeLuceneDocIdCollector(docIDs);
- IndexSearcher indexSearcher = null;
- try {
- Query query = new QueryParser(_column, _analyzer).parse(searchQuery);
- indexSearcher = _searcherManager.acquire();
- indexSearcher.search(query, docIDCollector);
- return getPinotDocIds(indexSearcher, docIDs);
- } catch (Exception e) {
- LOGGER
- .error("Failed while searching the realtime text index for column
{}, search query {}, exception {}", _column,
- searchQuery, e.getMessage());
- throw new RuntimeException(e);
- } finally {
+ RealtimeLuceneDocIdCollector docIDCollector = new
RealtimeLuceneDocIdCollector(docIDs);
+ // A thread interrupt during indexSearcher.search() can break the
underlying FSDirectory used by the IndexWriter
+ // which the SearcherManager is created with. To ensure the index is never
corrupted the search is executed
+ // in a child thread and the interrupt is handled in the current thread by
canceling the search gracefully.
+ // See https://github.com/apache/lucene/issues/3315 and
https://github.com/apache/lucene/issues/9309
+ Callable<MutableRoaringBitmap> searchCallable = () -> {
+ IndexSearcher indexSearcher = null;
try {
- if (indexSearcher != null) {
- _searcherManager.release(indexSearcher);
+ Query query = new QueryParser(_column, _analyzer).parse(searchQuery);
+ indexSearcher = _searcherManager.acquire();
+ indexSearcher.search(query, docIDCollector);
+ return getPinotDocIds(indexSearcher, docIDs);
+ } finally {
+ try {
+ if (indexSearcher != null) {
+ _searcherManager.release(indexSearcher);
+ }
+ } catch (Exception e) {
+ LOGGER.error(
+ "Failed while releasing the searcher manager for realtime text
index for column {}, exception {}",
+ _column, e.getMessage());
}
- } catch (Exception e) {
- LOGGER.error("Failed while releasing the searcher manager for realtime
text index for column {}, exception {}",
- _column, e.getMessage());
}
+ };
+ Future<MutableRoaringBitmap> searchFuture =
+ SEARCHER_POOL.getExecutorService().submit(searchCallable);
+ try {
+ return searchFuture.get();
+ } catch (InterruptedException e) {
+ docIDCollector.markShouldCancel();
Review Comment:
Just for clarification - does doing this prevent index corruption like
mentioned in PR description ? This is still in the context of searcher thread
that got interrupted right ?
--
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: commits-unsubscr...@pinot.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For additional commands, e-mail: commits-h...@pinot.apache.org
