honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Not load the ids of child documents into memory. 

Closes #3028
This commit is contained in:
Martijn van Groningen 2013-05-10 10:30:25 +02:00
parent 31b4b7ea58
commit 669cf90d0c
3 changed files with 21 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
src/main/java/org/elasticsearch/index/cache/id/IdCache.java vendored
View File

@ -29,7 +29,8 @@ import org.elasticsearch.index.service.IndexService;
import java.util.List;
/**
*
* This id cache contains only the ids of parent documents, loaded via the uid or parent field.
* This name IdCache is misleading, parentIdCache would be a better name.
*/
public interface IdCache extends IndexComponent, CloseableComponent, Iterable<IdReaderCache> {

19
src/main/java/org/elasticsearch/index/cache/id/simple/SimpleIdCache.java vendored
View File

@ -116,6 +116,15 @@ public class SimpleIdCache extends AbstractIndexComponent implements IdCache, Se
Map<Object, Map<String, TypeBuilder>> builders = new HashMap<Object, Map<String, TypeBuilder>>();
Map<Object, IndexReader> cacheToReader = new HashMap<Object, IndexReader>();
// We don't want to load uid of child documents, this allows us to not load uids of child types.
Set<HashedBytesArray> parentTypes = new HashSet<HashedBytesArray>();
for (String type : indexService.mapperService().types()) {
ParentFieldMapper parentFieldMapper = indexService.mapperService().documentMapper(type).parentFieldMapper();
if (parentFieldMapper != null) {
parentTypes.add(new HashedBytesArray(parentFieldMapper.type()));
}
}
// first, go over and load all the id->doc map for all types
for (AtomicReaderContext context : atomicReaderContexts) {
AtomicReader reader = context.reader();
@ -138,10 +147,16 @@ public class SimpleIdCache extends AbstractIndexComponent implements IdCache, Se
DocsEnum docsEnum = null;
for (BytesRef term = termsEnum.next(); term != null; term = termsEnum.next()) {
HashedBytesArray[] typeAndId = Uid.splitUidIntoTypeAndId(term);
TypeBuilder typeBuilder = readerBuilder.get(typeAndId[0].toUtf8());
// TODO: seek!
if (!parentTypes.contains(typeAndId[0])) {
continue;
}
String type = typeAndId[0].toUtf8();
TypeBuilder typeBuilder = readerBuilder.get(type);
if (typeBuilder == null) {
typeBuilder = new TypeBuilder(reader);
readerBuilder.put(typeAndId[0].toUtf8(), typeBuilder);
readerBuilder.put(type, typeBuilder);
}
HashedBytesArray idAsBytes = checkIfCanReuse(builders, typeAndId[1]);

3
src/test/java/org/elasticsearch/benchmark/search/child/ChildSearchBenchmark.java
View File

@ -69,7 +69,7 @@ public class ChildSearchBenchmark {
int QUERY_COUNT = 50;
String indexName = "test";
Thread.sleep(10000);
client.admin().cluster().prepareHealth(indexName).setWaitForGreenStatus().setTimeout("10s").execute().actionGet();
try {
client.admin().indices().create(createIndexRequest(indexName)).actionGet();
client.admin().indices().preparePutMapping(indexName).setType("child").setSource(XContentFactory.jsonBuilder().startObject().startObject("type")
@ -366,6 +366,7 @@ public class ChildSearchBenchmark {
System.out.println("--> has_parent query with match_all Query Avg: " + (totalQueryTime / QUERY_COUNT) + "ms");
System.gc();
statsResponse = client.admin().cluster().prepareNodesStats()
.setJvm(true).setIndices(true).execute().actionGet();