SOLR-3755: basic shard splitting code

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1379195 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Yonik Seeley 2012-08-30 22:31:56 +00:00
parent cfef80af55
commit 7754d1d105
9 changed files with 372 additions and 26 deletions

View File

@ -19,10 +19,13 @@ package org.apache.solr.handler.admin;
import java.io.File;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.Date;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Properties;
@ -36,6 +39,7 @@ import org.apache.solr.cloud.ZkController;
import org.apache.solr.common.SolrException;
import org.apache.solr.common.SolrException.ErrorCode;
import org.apache.solr.common.cloud.ClusterState;
import org.apache.solr.common.cloud.HashPartitioner;
import org.apache.solr.common.cloud.Slice;
import org.apache.solr.common.cloud.ZkNodeProps;
import org.apache.solr.common.cloud.ZkStateReader;
@ -57,6 +61,7 @@ import org.apache.solr.request.SolrQueryRequest;
import org.apache.solr.response.SolrQueryResponse;
import org.apache.solr.search.SolrIndexSearcher;
import org.apache.solr.update.MergeIndexesCommand;
import org.apache.solr.update.SplitIndexCommand;
import org.apache.solr.update.processor.UpdateRequestProcessor;
import org.apache.solr.update.processor.UpdateRequestProcessorChain;
import org.apache.solr.util.NumberUtils;
@ -171,6 +176,11 @@ public class CoreAdminHandler extends RequestHandlerBase {
break;
}
case SPLIT: {
doPersist = this.handleSplitAction(req, rsp);
break;
}
case PREPRECOVERY: {
this.handleWaitForStateAction(req, rsp);
break;
@ -202,6 +212,62 @@ public class CoreAdminHandler extends RequestHandlerBase {
rsp.setHttpCaching(false);
}
protected boolean handleSplitAction(SolrQueryRequest adminReq, SolrQueryResponse rsp) throws IOException {
SolrParams params = adminReq.getParams();
// partitions=N (split into N partitions, leaving it up to solr what the ranges are and where to put them)
// path - multiValued param, or comma separated param? Only creates indexes, not cores
List<HashPartitioner.Range> ranges = null;
// boolean closeDirectories = true;
// DirectoryFactory dirFactory = null;
String cname = params.get(CoreAdminParams.CORE, "");
SolrCore core = coreContainer.getCore(cname);
SolrQueryRequest req = new LocalSolrQueryRequest(core, params);
try {
String[] pathsArr = params.getParams("path");
List<String> paths = null;
String rangesStr = params.get("ranges"); // ranges=a-b,c-d,e-f
// dirFactory = core.getDirectoryFactory();
if (pathsArr != null) {
paths = Arrays.asList(pathsArr);
if (rangesStr == null) {
HashPartitioner hp = new HashPartitioner();
// should this be static?
// TODO: use real range if we know it. If we don't know it, we should prob
// split on every other doc rather than on a hash?
ranges = hp.partitionRange(pathsArr.length, Integer.MIN_VALUE, Integer.MAX_VALUE);
}
}
SplitIndexCommand cmd = new SplitIndexCommand(req, paths, ranges);
core.getUpdateHandler().split(cmd);
} catch (Exception e) {
log.error("ERROR executing split:", e);
throw new RuntimeException(e);
} finally {
if (req != null) req.close();
if (core != null) core.close();
}
return false;
}
protected boolean handleMergeAction(SolrQueryRequest req, SolrQueryResponse rsp) throws IOException {
SolrParams params = req.getParams();
String cname = params.required().get(CoreAdminParams.CORE);

View File

@ -738,6 +738,13 @@ public class DirectUpdateHandler2 extends UpdateHandler implements SolrCoreState
}
}
@Override
public void split(SplitIndexCommand cmd) throws IOException {
// TODO: do a commit first?
SolrIndexSplitter splitter = new SolrIndexSplitter(cmd);
splitter.split();
}
/////////////////////////////////////////////////////////////////////
// SolrInfoMBean stuff: Statistics and Module Info
/////////////////////////////////////////////////////////////////////

View File

@ -0,0 +1,183 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.solr.update;
import org.apache.lucene.index.AtomicReader;
import org.apache.lucene.index.AtomicReaderContext;
import org.apache.lucene.index.DocsEnum;
import org.apache.lucene.index.Fields;
import org.apache.lucene.index.FilterAtomicReader;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.Terms;
import org.apache.lucene.index.TermsEnum;
import org.apache.lucene.util.Bits;
import org.apache.lucene.util.BytesRef;
import org.apache.lucene.util.IOUtils;
import org.apache.lucene.util.OpenBitSet;
import org.apache.solr.common.cloud.HashPartitioner;
import org.apache.solr.common.util.Hash;
import org.apache.solr.core.SolrCore;
import org.apache.solr.schema.SchemaField;
import org.apache.solr.search.SolrIndexSearcher;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
public class SolrIndexSplitter {
public static Logger log = LoggerFactory.getLogger(SolrIndexSplitter.class);
SolrIndexSearcher searcher;
SchemaField field;
List<HashPartitioner.Range> ranges;
HashPartitioner.Range[] rangesArr; // same as ranges list, but an array for extra speed in inner loops
List<String> paths;
public SolrIndexSplitter(SplitIndexCommand cmd) {
field = cmd.getReq().getSchema().getUniqueKeyField();
searcher = cmd.getReq().getSearcher();
ranges = cmd.ranges;
rangesArr = ranges.toArray(new HashPartitioner.Range[ranges.size()]);
paths = cmd.paths;
}
public void split() throws IOException {
List<AtomicReaderContext> leaves = searcher.getTopReaderContext().leaves();
List<OpenBitSet[]> segmentDocSets = new ArrayList<OpenBitSet[]>(leaves.size());
log.info("SolrIndexSplitter: partitions=" + ranges.size() + " segments="+leaves.size());
for (AtomicReaderContext readerContext : leaves) {
assert readerContext.ordInParent == segmentDocSets.size(); // make sure we're going in order
OpenBitSet[] docSets = split(readerContext);
segmentDocSets.add( docSets );
}
// would it be more efficient to write segment-at-a-time to each new index?
// - need to worry about number of open descriptors
// - need to worry about if IW.addIndexes does a sync or not...
IndexReader[] subReaders = new IndexReader[leaves.size()];
for (int partitionNumber=0; partitionNumber<ranges.size(); partitionNumber++) {
log.info("SolrIndexSplitter: partition #" + partitionNumber + " range=" + ranges.get(partitionNumber));
for (int segmentNumber = 0; segmentNumber<subReaders.length; segmentNumber++) {
subReaders[segmentNumber] = new LiveDocsReader( leaves.get(segmentNumber), segmentDocSets.get(segmentNumber)[partitionNumber] );
}
String path = paths.get(partitionNumber);
boolean success = false;
SolrCore core = searcher.getCore();
IndexWriter iw = new SolrIndexWriter("SplittingIndexWriter"+partitionNumber + " " + ranges.get(partitionNumber), path,
core.getDirectoryFactory(), true, core.getSchema(),
core.getSolrConfig().indexConfig, core.getDeletionPolicy(), core.getCodec(), true);
try {
iw.addIndexes(subReaders);
// TODO: will many deletes have been removed, or should we optimize?
success = true;
} finally {
if (success) {
IOUtils.close(iw);
} else {
IOUtils.closeWhileHandlingException(iw);
}
}
}
}
OpenBitSet[] split(AtomicReaderContext readerContext) throws IOException {
AtomicReader reader = readerContext.reader();
OpenBitSet[] docSets = new OpenBitSet[ranges.size()];
for (int i=0; i<docSets.length; i++) {
docSets[i] = new OpenBitSet(reader.maxDoc());
}
Bits liveDocs = reader.getLiveDocs();
Fields fields = reader.fields();
Terms terms = fields==null ? null : fields.terms(field.getName());
TermsEnum termsEnum = terms==null ? null : terms.iterator(null);
if (termsEnum == null) return docSets;
BytesRef term = null;
DocsEnum docsEnum = null;
for (;;) {
term = termsEnum.next();
if (term == null) break;
// figure out the hash for the term
// TODO: hook in custom hashes (or store hashes)
int hash = Hash.murmurhash3_x86_32(term.bytes, term.offset, term.length, 0);
docsEnum = termsEnum.docs(liveDocs, docsEnum, 0x0);
for (;;) {
int doc = docsEnum.nextDoc();
if (doc == DocsEnum.NO_MORE_DOCS) break;
for (int i=0; i<rangesArr.length; i++) { // inner-loop: use array here for extra speed.
if (rangesArr[i].includes(hash)) {
docSets[i].fastSet(doc);
}
}
}
}
return docSets;
}
// change livedocs on the reader to delete those docs we don't want
static class LiveDocsReader extends FilterAtomicReader {
final OpenBitSet liveDocs;
final int numDocs;
public LiveDocsReader(AtomicReaderContext context, OpenBitSet liveDocs) throws IOException {
super(context.reader());
this.liveDocs = liveDocs;
this.numDocs = (int)liveDocs.cardinality();
}
@Override
public int numDocs() {
return numDocs;
}
@Override
public boolean hasDeletions() {
return (in.maxDoc() != numDocs);
}
@Override
public Bits getLiveDocs() {
return liveDocs;
}
}
}

View File

@ -0,0 +1,56 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.solr.update;
import org.apache.solr.common.cloud.HashPartitioner;
import org.apache.solr.request.SolrQueryRequest;
import java.util.List;
/**
* A merge indexes command encapsulated in an object.
*
* @since solr 1.4
*
*/
public class SplitIndexCommand extends UpdateCommand {
// public List<Directory> dirs;
public List<String> paths;
public List<HashPartitioner.Range> ranges;
// TODO: allow specification of custom hash function
public SplitIndexCommand(SolrQueryRequest req, List<String> paths, List<HashPartitioner.Range> ranges) {
super(req);
this.paths = paths;
this.ranges = ranges;
}
@Override
public String name() {
return "split";
}
@Override
public String toString() {
StringBuilder sb = new StringBuilder(super.toString());
sb.append(",paths=" + paths);
sb.append(",ranges=" + ranges);
sb.append('}');
return sb.toString();
}
}

View File

@ -179,4 +179,6 @@ public abstract class UpdateHandler implements SolrInfoMBean {
{
optimizeCallbacks.add( listener );
}
public abstract void split(SplitIndexCommand cmd) throws IOException;
}

View File

@ -27,12 +27,25 @@ public class TestHashPartitioner extends SolrTestCaseJ4 {
public void testMapHashes() throws Exception {
HashPartitioner hp = new HashPartitioner();
for (int i = 1; i <= 30000; i++) {
List<Range> ranges = hp.partitionRange(i);
List<Range> ranges;
// make sure the partitioner uses the "natural" boundaries and doesn't suffer from an off-by-one
ranges = hp.partitionRange(2, Integer.MIN_VALUE, Integer.MAX_VALUE);
assertEquals(Integer.MIN_VALUE, ranges.get(0).min);
assertEquals(0x80000000, ranges.get(0).min);
assertEquals(0xffffffff, ranges.get(0).max);
assertEquals(0x00000000, ranges.get(1).min);
assertEquals(0x7fffffff, ranges.get(1).max);
ranges = hp.partitionRange(2, 0, 0x7fffffff);
assertEquals(0x00000000, ranges.get(0).min);
assertEquals(0x3fffffff, ranges.get(0).max);
assertEquals(0x40000000, ranges.get(1).min);
assertEquals(0x7fffffff, ranges.get(1).max);
for (int i = 1; i <= 30000; i += 13) {
ranges = hp.partitionRange(i, Integer.MIN_VALUE, Integer.MAX_VALUE);
assertEquals(i, ranges.size());
assertTrue("First range does not start before " + Integer.MIN_VALUE
+ " it is:" + ranges.get(0).min,
ranges.get(0).min <= Integer.MIN_VALUE);

View File

@ -226,7 +226,7 @@ public class ClusterState implements JSONWriter.Writable {
shardList.addAll(shards);
Collections.sort(shardList);
ranges = hp.partitionRange(shards.size());
ranges = hp.partitionRange(shards.size(), Integer.MIN_VALUE, Integer.MAX_VALUE);
rangeInfo.ranges = ranges;
rangeInfo.shardList = shardList;
@ -243,7 +243,7 @@ public class ClusterState implements JSONWriter.Writable {
int cnt = 0;
for (Range range : rangInfo.ranges) {
if (hash < range.max) {
if (range.includes(hash)) {
return rangInfo.shardList.get(cnt);
}
cnt++;

View File

@ -25,39 +25,57 @@ import java.util.List;
*
*/
public class HashPartitioner {
// Hash ranges can't currently "wrap" - i.e. max must be greater or equal to min.
public static class Range {
public long min;
public long max;
public int min; // inclusive
public int max; // inclusive
public Range(long min, long max) {
public Range(int min, int max) {
this.min = min;
this.max = max;
}
public boolean includes(int hash) {
return hash >= min && hash <= max;
}
public String toString() {
return Integer.toHexString(min) + '-' + Integer.toHexString(max);
}
public static Range fromString(String range) {
return null; // TODO
}
}
/**
* works up to 65537 before requested num of ranges is one short
*
*
* @param partitions
* @return Range for each partition
*/
public List<Range> partitionRange(int partitions) {
// some hokey code to partition the int space
long range = Integer.MAX_VALUE + (Math.abs((long) Integer.MIN_VALUE));
long srange = range / partitions;
public List<Range> partitionRange(int partitions, int min, int max) {
assert max >= min;
long range = (long)max - (long)min;
long srange = Math.max(1, range / partitions);
List<Range> ranges = new ArrayList<Range>(partitions);
long end = 0;
long start = Integer.MIN_VALUE;
while (end < Integer.MAX_VALUE) {
long start = min;
long end = start;
while (end < max) {
end = start + srange;
ranges.add(new Range(start, end));
// make last range always end exactly on MAX_VALUE
if (ranges.size() == partitions - 1) {
end = max;
}
ranges.add(new Range((int)start, (int)end));
start = end + 1L;
}
return ranges;
}
}

View File

@ -92,7 +92,8 @@ public interface CoreAdminParams
SWAP,
RENAME,
MERGEINDEXES,
PREPRECOVERY,
SPLIT,
PREPRECOVERY,
REQUESTRECOVERY,
REQUESTSYNCSHARD;