open PdbReaders only when reading

We used to open all PdbReaders in a search result and then interate over
them. This used a lot of heap space (> 8GB) for 400k files.
Now the PdbReaders are only opened while they are used. Heap usage was
less than 550 while reading more than 400k files.
This commit is contained in:
2017-11-18 10:12:22 +01:00
parent a636f2b9bd
commit cc49a8cf2a
4 changed files with 253 additions and 238 deletions

View File

@@ -7,23 +7,23 @@ import org.lucares.pdb.api.Tags;
class Group {
private final Tags tags;
private final List<PdbReader> readers;
private final List<PdbFile> files;
public Group(final Tags tags, final List<PdbReader> files) {
public Group(final Tags tags, final List<PdbFile> files) {
super();
this.tags = tags;
this.readers = files;
this.files = files;
}
public Tags getTags() {
return tags;
}
public List<PdbReader> getReaders() {
return readers;
public List<PdbFile> getFiles() {
return files;
}
public void addReader(final PdbReader pdbReader) {
readers.add(pdbReader);
public void addFile(final PdbFile file) {
files.add(file);
}
}

View File

@@ -23,22 +23,22 @@ public class Grouping {
this.groups.addAll(groups);
}
public static Grouping groupBy(final List<PdbReader> pdbReaders, final List<String> groupByField) {
public static Grouping groupBy(final List<PdbFile> pdbFiles, final List<String> groupByField) {
final Grouping result;
if (noGrouping(groupByField)) {
final Group group = new Group(Tags.EMPTY, pdbReaders);
final Group group = new Group(Tags.EMPTY, pdbFiles);
result = new Grouping(group);
} else {
final Map<Tags, Group> grouping = new HashMap<>();
for (final PdbReader pdbReader : pdbReaders) {
final Tags tags = pdbReader.getPdbFile().getTags();
for (final PdbFile pdbFile : pdbFiles) {
final Tags tags = pdbFile.getTags();
final Tags groupTags = tags.subset(groupByField);
addIfNotExists(grouping, groupTags);
grouping.get(groupTags).addReader(pdbReader);
grouping.get(groupTags).addFile(pdbFile);
}
result = new Grouping(grouping.values());
}
@@ -51,9 +51,9 @@ public class Grouping {
private static void addIfNotExists(final Map<Tags, Group> grouping, final Tags groupTags) {
if (!grouping.containsKey(groupTags)) {
final List<PdbReader> readers = new ArrayList<>();
final List<PdbFile> files = new ArrayList<>();
grouping.put(groupTags, new Group(groupTags, readers));
grouping.put(groupTags, new Group(groupTags, files));
}
}

View File

@@ -1,5 +1,8 @@
package org.lucares.performance.db;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.nio.file.Files;
import java.util.ArrayDeque;
import java.util.Collection;
import java.util.Iterator;
@@ -13,14 +16,17 @@ import org.slf4j.LoggerFactory;
public class PdbFileIterator implements Iterator<Entry>, AutoCloseable {
private final static Logger LOGGER = LoggerFactory.getLogger(PdbFileIterator.class);
private final static Logger LOGGER = LoggerFactory
.getLogger(PdbFileIterator.class);
private static final class EntrySupplier implements Supplier<Entry>, AutoCloseable {
private static final class EntrySupplier implements Supplier<Entry>,
AutoCloseable {
private final Queue<PdbReader> pdbFiles;
private final Queue<PdbFile> pdbFiles;
private PdbReader reader;
private PdbFile currentPdbFile;
public EntrySupplier(final Collection<PdbReader> pdbFiles) {
public EntrySupplier(final Collection<PdbFile> pdbFiles) {
super();
this.pdbFiles = new ArrayDeque<>(pdbFiles);
}
@@ -45,7 +51,8 @@ public class PdbFileIterator implements Iterator<Entry>, AutoCloseable {
// A reader might return null, for a newly opened reader,
// if the file was created, but nothing has been written to
// disk yet.
// This might happen, because of buffering, or when an ingestion
// This might happen, because of buffering, or when an
// ingestion
// was cancelled.
}
}
@@ -61,7 +68,23 @@ public class PdbFileIterator implements Iterator<Entry>, AutoCloseable {
reader = null;
}
reader = pdbFiles.poll();
while (!pdbFiles.isEmpty()) {
currentPdbFile = pdbFiles.poll();
try {
if (Files.size(currentPdbFile.getPath()) > 0) {
reader = new PdbReader(currentPdbFile);
break;
} else {
LOGGER.info("ignoring empty file " + currentPdbFile);
}
} catch (final FileNotFoundException e) {
LOGGER.warn("the pdbFile " + currentPdbFile.getPath()
+ " is missing", e);
} catch (final IOException e) {
throw new ReadException(e);
}
}
}
@Override
@@ -69,14 +92,6 @@ public class PdbFileIterator implements Iterator<Entry>, AutoCloseable {
if (reader != null) {
reader.close();
}
while (!pdbFiles.isEmpty()) {
try {
pdbFiles.poll().close();
} catch (final Exception e) {
LOGGER.warn("Closing pdb file failed.", e);
}
}
}
}
@@ -85,7 +100,7 @@ public class PdbFileIterator implements Iterator<Entry>, AutoCloseable {
private Optional<Entry> next = Optional.empty();
public PdbFileIterator(final Collection<PdbReader> pdbFiles) {
public PdbFileIterator(final Collection<PdbFile> pdbFiles) {
supplier = new EntrySupplier(pdbFiles);
}

View File

@@ -151,9 +151,9 @@ public class PerformanceDb implements AutoCloseable {
*/
public Result get(final String query, final List<String> groupBy) {
final List<PdbReader> pdbReaders = tagsToFile.getReaders(query);
final List<PdbFile> pdbFiles = tagsToFile.getFilesForQuery(query);
final Grouping grouping = Grouping.groupBy(pdbReaders, groupBy);
final Grouping grouping = Grouping.groupBy(pdbFiles, groupBy);
final Result result = toResult(grouping);
@@ -163,7 +163,7 @@ public class PerformanceDb implements AutoCloseable {
private Result toResult(final Grouping grouping) {
final List<GroupResult> groupResults = new ArrayList<>();
for (final Group group : grouping.getGroups()) {
final Stream<Entry> stream = toStream(group.getReaders());
final Stream<Entry> stream = toStream(group.getFiles());
final GroupResult groupResult = new GroupResult(stream, group.getTags());
groupResults.add(groupResult);
}
@@ -171,7 +171,7 @@ public class PerformanceDb implements AutoCloseable {
return result;
}
private Stream<Entry> toStream(final List<PdbReader> pdbFiles) {
private Stream<Entry> toStream(final List<PdbFile> pdbFiles) {
final PdbFileIterator iterator = new PdbFileIterator(pdbFiles);
final Spliterator<Entry> spliterator = Spliterators.spliteratorUnknownSize(iterator, Spliterator.ORDERED);