Apache
/
lucene
mirror of https://github.com/apache/lucene.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

SOLR-5353: Enhance CoreAdmin api to split a route key's documents from an index and leave behind all other documents 

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1533163 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Shalin Shekhar Mangar 2013-10-17 17:08:59 +00:00
parent 8b47e1c575
commit ea81f1add6
6 changed files with 109 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
solr/CHANGES.txt
View File

@ -97,6 +97,9 @@ New Features
* SOLR-5338: Split shards by a route key using split.key parameter. (shalin)
* SOLR-5353: Enhance CoreAdmin api to split a route key's documents from an index
and leave behind all other documents. (shalin)
Bug Fixes
----------------------

3
solr/core/src/java/org/apache/solr/handler/admin/CoreAdminHandler.java
View File

@ -249,6 +249,7 @@ public class CoreAdminHandler extends RequestHandlerBase {
}
}
}
String splitKey = params.get("split.key");
String[] newCoreNames = params.getParams("targetCore");
String cname = params.get(CoreAdminParams.CORE, "");
@ -300,7 +301,7 @@ public class CoreAdminHandler extends RequestHandlerBase {
}
SplitIndexCommand cmd = new SplitIndexCommand(req, paths, newCores, ranges, router, routeFieldName);
SplitIndexCommand cmd = new SplitIndexCommand(req, paths, newCores, ranges, router, routeFieldName, splitKey);
core.getUpdateHandler().split(cmd);
// After the split has completed, someone (here?) should start the process of replaying the buffered updates.

32
solr/core/src/java/org/apache/solr/update/SolrIndexSplitter.java
View File

@ -31,6 +31,7 @@ import org.apache.lucene.util.BytesRef;
import org.apache.lucene.util.CharsRef;
import org.apache.lucene.util.IOUtils;
import org.apache.lucene.util.OpenBitSet;
import org.apache.solr.common.cloud.CompositeIdRouter;
import org.apache.solr.common.cloud.DocRouter;
import org.apache.solr.common.cloud.HashBasedRouter;
import org.apache.solr.core.SolrCore;
@ -58,6 +59,7 @@ public class SolrIndexSplitter {
int numPieces;
int currPartition = 0;
String routeFieldName;
String splitKey;
public SolrIndexSplitter(SplitIndexCommand cmd) {
searcher = cmd.getReq().getSearcher();
@ -79,6 +81,9 @@ public class SolrIndexSplitter {
} else {
field = searcher.getSchema().getField(routeFieldName);
}
if (cmd.splitKey != null) {
splitKey = getRouteKey(cmd.splitKey);
}
}
public void split() throws IOException {
@ -174,11 +179,20 @@ public class SolrIndexSplitter {
idRef = field.getType().indexedToReadable(term, idRef);
String idString = idRef.toString();
if (splitKey != null) {
// todo have composite routers support these kind of things instead
String part1 = getRouteKey(idString);
if (part1 == null)
continue;
if (!splitKey.equals(part1)) {
continue;
}
}
int hash = 0;
if (hashRouter != null) {
hash = hashRouter.sliceHash(idString, null, null, null);
}
// int hash = Hash.murmurhash3_x86_32(ref, ref.offset, ref.length, 0);
docsEnum = termsEnum.docs(liveDocs, docsEnum, DocsEnum.FLAG_NONE);
for (;;) {
@ -200,6 +214,22 @@ public class SolrIndexSplitter {
return docSets;
}
private String getRouteKey(String idString) {
int idx = idString.indexOf(CompositeIdRouter.separator);
if (idx <= 0) return null;
String part1 = idString.substring(0, idx);
int commaIdx = part1.indexOf(CompositeIdRouter.bitsSeparator);
if (commaIdx > 0) {
if (commaIdx + 1 < part1.length()) {
char ch = part1.charAt(commaIdx + 1);
if (ch >= '0' && ch <= '9') {
part1 = part1.substring(0, commaIdx);
}
}
}
return part1;
}
// change livedocs on the reader to delete those docs we don't want
static class LiveDocsReader extends FilterAtomicReader {

10
solr/core/src/java/org/apache/solr/update/SplitIndexCommand.java
View File

@ -36,14 +36,16 @@ public class SplitIndexCommand extends UpdateCommand {
public List<DocRouter.Range> ranges;
public DocRouter router;
public String routeFieldName;
public String splitKey;
public SplitIndexCommand(SolrQueryRequest req, List<String> paths, List<SolrCore> cores, List<DocRouter.Range> ranges, DocRouter router, String routeFieldName) {
public SplitIndexCommand(SolrQueryRequest req, List<String> paths, List<SolrCore> cores, List<DocRouter.Range> ranges, DocRouter router, String routeFieldName, String splitKey) {
super(req);
this.paths = paths;
this.cores = cores;
this.ranges = ranges;
this.router = router;
this.routeFieldName = routeFieldName;
this.splitKey = splitKey;
}
@Override
@ -58,6 +60,12 @@ public class SplitIndexCommand extends UpdateCommand {
sb.append(",cores=" + cores);
sb.append(",ranges=" + ranges);
sb.append(",router=" + router);
if (routeFieldName != null) {
sb.append(",routeFieldName=" + routeFieldName);
}
if (splitKey != null) {
sb.append(",split.key=" + splitKey);
}
sb.append('}');
return sb.toString();
}

66
solr/core/src/test/org/apache/solr/update/SolrIndexSplitterTest.java
View File

@ -25,6 +25,7 @@ import org.apache.lucene.store.Directory;
import org.apache.solr.SolrTestCaseJ4;
import org.apache.solr.client.solrj.SolrQuery;
import org.apache.solr.client.solrj.embedded.EmbeddedSolrServer;
import org.apache.solr.common.cloud.CompositeIdRouter;
import org.apache.solr.common.cloud.DocRouter;
import org.apache.solr.common.cloud.PlainIdRouter;
import org.apache.solr.common.util.Hash;
@ -96,7 +97,7 @@ public class SolrIndexSplitterTest extends SolrTestCaseJ4 {
request = lrf.makeRequest("q", "dummy");
SplitIndexCommand command = new SplitIndexCommand(request,
Lists.newArrayList(indexDir1.getAbsolutePath(), indexDir2.getAbsolutePath()), null, ranges, new PlainIdRouter(), null);
Lists.newArrayList(indexDir1.getAbsolutePath(), indexDir2.getAbsolutePath()), null, ranges, new PlainIdRouter(), null, null);
new SolrIndexSplitter(command).split();
Directory directory = h.getCore().getDirectoryFactory().get(indexDir1.getAbsolutePath(),
@ -141,7 +142,7 @@ public class SolrIndexSplitterTest extends SolrTestCaseJ4 {
request = lrf.makeRequest("q", "dummy");
SplitIndexCommand command = new SplitIndexCommand(request,
Lists.newArrayList(indexDir1.getAbsolutePath(), indexDir2.getAbsolutePath()), null, ranges, new PlainIdRouter(), null);
Lists.newArrayList(indexDir1.getAbsolutePath(), indexDir2.getAbsolutePath()), null, ranges, new PlainIdRouter(), null, null);
new SolrIndexSplitter(command).split();
Directory directory = h.getCore().getDirectoryFactory().get(indexDir1.getAbsolutePath(),
@ -198,7 +199,7 @@ public class SolrIndexSplitterTest extends SolrTestCaseJ4 {
try {
request = lrf.makeRequest("q", "dummy");
SplitIndexCommand command = new SplitIndexCommand(request, null, Lists.newArrayList(core1, core2), ranges, new PlainIdRouter(), null);
SplitIndexCommand command = new SplitIndexCommand(request, null, Lists.newArrayList(core1, core2), ranges, new PlainIdRouter(), null, null);
new SolrIndexSplitter(command).split();
} finally {
if (request != null) request.close();
@ -235,7 +236,7 @@ public class SolrIndexSplitterTest extends SolrTestCaseJ4 {
request = lrf.makeRequest("q", "dummy");
SplitIndexCommand command = new SplitIndexCommand(request,
Lists.newArrayList(indexDir1.getAbsolutePath(), indexDir2.getAbsolutePath(), indexDir3.getAbsolutePath()), null, null, new PlainIdRouter(), null);
Lists.newArrayList(indexDir1.getAbsolutePath(), indexDir2.getAbsolutePath(), indexDir3.getAbsolutePath()), null, null, new PlainIdRouter(), null, null);
new SolrIndexSplitter(command).split();
directory = h.getCore().getDirectoryFactory().get(indexDir1.getAbsolutePath(),
@ -266,6 +267,63 @@ public class SolrIndexSplitterTest extends SolrTestCaseJ4 {
}
}
@Test
public void testSplitByRouteKey() throws Exception {
File indexDir = new File(TEMP_DIR, this.getClass().getName() + "testSplitByRouteKey");
if (indexDir.exists()) {
FileUtils.deleteDirectory(indexDir);
}
indexDir.mkdirs();
CompositeIdRouter r1 = new CompositeIdRouter();
String splitKey = "sea-line!";
String key2 = "soul-raising!";
// murmur2 has a collision on the above two keys
assertEquals(r1.keyHashRange(splitKey), r1.keyHashRange(key2));
/*
More strings with collisions on murmur2 for future reference:
"Drava" "dessert spoon"
"Bighorn" "pleasure lover"
"attributable to" "second edition"
"sea-line" "soul-raising"
"lift direction" "testimony meeting"
*/
for (int i=0; i<10; i++) {
assertU(adoc("id", splitKey + i));
assertU(adoc("id", key2 + i));
}
assertU(commit());
assertJQ(req("q", "*:*"), "/response/numFound==20");
DocRouter.Range splitKeyRange = r1.keyHashRange(splitKey);
LocalSolrQueryRequest request = null;
Directory directory = null;
try {
request = lrf.makeRequest("q", "dummy");
SplitIndexCommand command = new SplitIndexCommand(request,
Lists.newArrayList(indexDir.getAbsolutePath()), null, Lists.newArrayList(splitKeyRange), new CompositeIdRouter(), null, splitKey);
new SolrIndexSplitter(command).split();
directory = h.getCore().getDirectoryFactory().get(indexDir.getAbsolutePath(),
DirectoryFactory.DirContext.DEFAULT, h.getCore().getSolrConfig().indexConfig.lockType);
DirectoryReader reader = DirectoryReader.open(directory);
assertEquals("split index has wrong number of documents", 10, reader.numDocs());
reader.close();
h.getCore().getDirectoryFactory().release(directory);
directory = null;
} finally {
if (request != null) {
request.close();
}
if (directory != null) {
h.getCore().getDirectoryFactory().release(directory);
}
}
}
private List<DocRouter.Range> getRanges(String id1, String id2) throws UnsupportedEncodingException {
// find minHash/maxHash hash ranges
byte[] bytes = id1.getBytes("UTF-8");

4
solr/solrj/src/java/org/apache/solr/common/cloud/CompositeIdRouter.java
View File

@ -34,10 +34,10 @@ import java.util.List;
public class CompositeIdRouter extends HashBasedRouter {
public static final String NAME = "compositeId";
private static final int separator = '!';
public static final int separator = '!';
// separator used to optionally specify number of bits to allocate toward first part.
private static final int bitsSeparator = '/';
public static final int bitsSeparator = '/';
private int bits = 16;
private int mask1 = 0xffff0000;
private int mask2 = 0x0000ffff;