我们使用Solr Replication可以实现Solr服务器的可用性，即使某一个索引副本由于磁盘介质故障或者误操作删除等，其他的多个复制副本仍然可以提供服务。如果只是单纯的基于Solr Replication技术，只能对一个索引进行管理维护，当索引数据达到一定规模，搜索的性能成了瓶颈，除了重新规划设计索引，实现逻辑划分以外，没有更好地方法实现查询服务器的可扩展性。
SolrCloud就是为了解决这个问题而提出的。SolrCloud通过ZooKeeper集群来进行协调，使一个索引（SolrCloud中叫做一个Collection）进行分片，各个分片可以分布在不同的物理节点上，而且，对于同一个Collection的多个分片（Shard）之间没有交集，亦即，多个物理分片组成一个完成的索引Collection。为了保证分片数据的可用性，SolrCloud自动支持Solr Replication，可以同时对分片进行复制，冗余存储。下面，我们基于Solr最新的4.3.1版本进行安装配置SolrCloud集群，通过实践来实现索引数据的分布存储和检索。
准备工作
- 服务器信息
三台服务器：
10.95.3.61 master 10.95.3.62 slave1 10.95.3.65 slave4
- ZooKeeper集群配置
安装ZooKeeper集群，在上面3分节点上分别安装，使用的版本是zookeeper-3.4.5。 首先，在master节点上配置zoo.cfg，内容如下所示：
[hadoop@master ~]$ vi applications/zookeeper/zookeeper-3.4.5/conf/zoo.cfg # The number of milliseconds of each tick tickTime=2000 # The number of ticks that the initial # synchronization phase can take initLimit=10 # The number of ticks that can pass between # sending a request and getting an acknowledgement syncLimit=5 # the directory where the snapshot is stored. # do not use /tmp for storage, /tmp here is just # example sakes. dataDir=/home/hadoop/applications/zookeeper/zookeeper-3.4.5/data # the port at which the clients will connect clientPort=2188 dataLogDir=/home/hadoop/applications/zookeeper/zookeeper-3.4.5/data/logs server.1=master:4888:5888 server.2=slave1:4888:5888 server.3=slave4:4888:5888 # # Be sure to read the maintenance section of the # administrator guide before turning on autopurge. # # http://zookeeper.apache.org/doc/current/zookeeperAdmin.html#sc_maintenance # # The number of snapshots to retain in dataDir #autopurge.snapRetainCount=3 # Purge task interval in hours # Set to "0" to disable auto purge feature #autopurge.purgeInterval=1
然后，创建对应的数据存储目录后，可以直接将该配置复制到其他两个节点上：
scp -r applications/zookeeper/zookeeper-3.4.5 hadoop@slave1:~/applications/zookeeper/ scp -r applications/zookeeper/zookeeper-3.4.5 hadoop@slave4:~/applications/zookeeper/
启动ZooKeeper集群，在每个节点上分别启动ZooKeeper服务：
cd applications/zookeeper/zookeeper-3.4.5/ bin/zkServer.sh start
可以查看ZooKeeper集群的状态，保证集群启动没有问题：
[hadoop@master zookeeper-3.4.5]$ bin/zkServer.sh status JMX enabled by default Using config: /home/hadoop/applications/zookeeper/zookeeper-3.4.5/bin/../conf/zoo.cfg Mode: follower [hadoop@slave1 zookeeper-3.4.5]$ bin/zkServer.sh status JMX enabled by default Using config: /home/hadoop/applications/zookeeper/zookeeper-3.4.5/bin/../conf/zoo.cfg Mode: follower [hadoop@slave4 zookeeper-3.4.5]$ bin/zkServer.sh status JMX enabled by default Using config: /home/hadoop/applications/zookeeper/zookeeper-3.4.5/bin/../conf/zoo.cfg Mode: leader
可以看到，slave4节点是ZooKeeper集群服务Leader。
- SolrCloud相关目录
我们选择/home/hadoop/applications/solr/cloud目录存放Solr的库文件和配置文件，该目录下有lib和multicore两个子目录。 另外，还有一个存储索引的目录，设置为/home/hadoop/applications/storage/cloud/data。
SolrCloud配置
首先在一个节点上对SOLR进行配置，我们选择master节点。
1、SOLR基本配置
将下载的SOLR的压缩包解压缩，将solr-4.3.1\example\webapps\solr.war解开，将solr-4.3.1\example\webapps\solr\WEB-INF\lib和solr-4.3.1\example\lib\ext中的jar文件拷贝到solr-4.3.1\example\webapps\solr\WEB-INF\lib中，并将解开的solr目录改名为solr-cloud，然后传到服务器的Tomcat下的webapps目录下。 将solr-4.3.1\example\webapps\solr\WEB-INF\lib和solr-4.3.1\example\lib\ext下面的jar文件都拷贝到指定目录/home/hadoop/applications/solr/cloud/lib/中：
[hadoop@master ~]$ ls /home/hadoop/applications/solr/cloud/lib/ commons-cli-1.2.jar lucene-analyzers-common-4.3.1.jar lucene-suggest-4.3.1.jar commons-codec-1.7.jar lucene-analyzers-kuromoji-4.3.1.jar noggit-0.5.jar commons-fileupload-1.2.1.jar lucene-analyzers-phonetic-4.3.1.jar org.restlet-2.1.1.jar commons-io-2.1.jar lucene-codecs-4.3.1.jar org.restlet.ext.servlet-2.1.1.jar commons-lang-2.6.jar lucene-core-4.3.1.jar slf4j-api-1.6.6.jar guava-13.0.1.jar lucene-grouping-4.3.1.jar slf4j-log4j12-1.6.6.jar httpclient-4.2.3.jar lucene-highlighter-4.3.1.jar solr-core-4.3.1.jar httpcore-4.2.2.jar lucene-memory-4.3.1.jar solr-solrj-4.3.1.jar httpmime-4.2.3.jar lucene-misc-4.3.1.jar spatial4j-0.3.jar jcl-over-slf4j-1.6.6.jar lucene-queries-4.3.1.jar wstx-asl-3.2.7.jar jul-to-slf4j-1.6.6.jar lucene-queryparser-4.3.1.jar zookeeper-3.4.5.jar log4j-1.2.16.jar lucene-spatial-4.3.1.jar
目录/home/hadoop/applications/solr/cloud/multicore的结构，如图所示：
下面，我们对上面conf目录下的配置文件进行说明：
- schema.xml文件
<?xml version="1.0" ?> <schema name="example core two" version="1.1"> <types> <fieldtype name="string" omitNorms="true" /> <fieldType name="long" /> <fieldtype name="int" /> <fieldtype name="float" /> <fieldType name="date" precisionStep="0" positionIncrementGap="0" /> </types> <fields> <field name="id" type="long" indexed="true" stored="true" multiValued="false" required="true" /> <field name="area" type="string" indexed="true" stored="false" multiValued="false" /> <field name="building_type" type="int" indexed="true" stored="false" multiValued="false" /> <field name="category" type="string" indexed="true" stored="false" multiValued="false" /> <field name="temperature" type="int" indexed="true" stored="false" multiValued="false" /> <field name="code" type="int" indexed="true" stored="false" multiValued="false" /> <field name="latitude" type="float" indexed="true" stored="false" multiValued="false" /> <field name="longitude" type="float" indexed="true" stored="false" multiValued="false" /> <field name="when" type="date" indexed="true" stored="false" multiValued="false" /> <field name="_version_" type="long" indexed="true" stored="true" /> </fields> <uniqueKey>id</uniqueKey> <defaultSearchField>area</defaultSearchField> <solrQueryParser defaultOperator="OR" /> </schema>
- solrconfig.xml文件
<?xml version="1.0" encoding="UTF-8" ?>
<config>
<luceneMatchVersion>LUCENE_43</luceneMatchVersion>
<directoryFactory name="DirectoryFactory" />
<dataDir>${solr.shard.data.dir:}</dataDir>
<schemaFactory />
<updateHandler>
<updateLog>
<str name="dir">${solr.shard.data.dir:}</str>
</updateLog>
</updateHandler>
<!-- realtime get handler, guaranteed to return the latest stored fields of any document, without the need to commit or open a new searcher. The current implementation relies on the updateLog feature
being enabled. -->
<requestHandler name="/get">
<lst name="defaults">
<str name="omitHeader">true</str>
</lst>
</requestHandler>
<requestHandler name="/replication" startup="lazy" />
<requestDispatcher handleSelect="true">
<requestParsers enableRemoteStreaming="false" multipartUploadLimitInKB="2048" formdataUploadLimitInKB="2048" />
</requestDispatcher>
<requestHandler name="standard" default="true" />
<requestHandler name="/analysis/field" startup="lazy" />
<requestHandler name="/update" />
<requestHandler name="/update/csv" startup="lazy">
<lst name="defaults">
<str name="separator">,</str>
<str name="header">true</str>
<str name="encapsulator">"</str>
</lst>
<updateLog>
<str name="dir">${solr.shard.data.dir:}</str>
</updateLog>
</requestHandler>
<requestHandler name="/admin/" />
<requestHandler name="/admin/ping">
<lst name="invariants">
<str name="q">solrpingquery</str>
</lst>
<lst name="defaults">
<str name="echoParams">all</str>
</lst>
</requestHandler>
<updateRequestProcessorChain name="sample">
<processor />
<processor />
<processor />
</updateRequestProcessorChain>
<query>
<maxBooleanClauses>1024</maxBooleanClauses>
<filterCache size="10240" initialSize="512" autowarmCount="0" />
<queryResultCache size="10240" initialSize="512" autowarmCount="0" />
<documentCache size="10240" initialSize="512" autowarmCount="0" />
<enableLazyFieldLoading>true</enableLazyFieldLoading>
<queryResultWindowSize>20</queryResultWindowSize>
<queryResultMaxDocsCached>200</queryResultMaxDocsCached>
<maxWarmingSearchers>2</maxWarmingSearchers>
</query>
<admin>
<defaultQuery>solr</defaultQuery>
</admin>
</config>
- solrcore.properties文件
solr.shard.data.dir=/home/hadoop/applications/storage/cloud/data
属性solr.shard.data.dir在solrconfig.xml文件中被引用过，指定索引数据的存放位置。
- solr.xml文件
该文件中指定了ZooKeeper的相关配置，已经Solr Core的配置内容：
<?xml version="1.0" encoding="UTF-8" ?>
<solr persistent="true">
<cores defaultCoreName="collection1" host="${host:}" adminPath="/admin/cores" zkClientTimeout="${zkClientTimeout:15000}" hostPort="8888" hostContext="${hostContext:solr-cloud}">
</cores>
</solr>
注意：这里，我们并没有配置任何的core元素，这个等到整个配置安装完成之后，通过SOLR提供的REST接口，来实现Collection以及Shard的创建，从而来更新这些配置文件。
2、ZooKeeper管理监控配置文件
SolrCloud是通过ZooKeeper集群来保证配置文件的变更及时同步到各个节点上，所以，需要将配置文件上传到ZooKeeper集群中：
java -classpath .:/home/hadoop/applications/solr/cloud/lib/* org.apache.solr.cloud.ZkCLI -cmd upconfig -zkhost master:2188,slave1:2188,slave4:2188 -confdir /home/hadoop/applications/solr/cloud/multicore/collection1/conf -confname myconf java -classpath .:/home/hadoop/applications/solr/cloud/lib/* org.apache.solr.cloud.ZkCLI -cmd linkconfig -collection collection1 -confname myconf -zkhost master:2188,slave1:2188,slave4:2188
上传完成以后，我们检查一下ZooKeeper上的存储情况：
[hadoop@master ~]$ cd applications/zookeeper/zookeeper-3.4.5/ [hadoop@master zookeeper-3.4.5]$ bin/zkCli.sh -server master:2188 ... [zk: master:2188(CONNECTED) 0] ls / [configs, collections, zookeeper] [zk: master:2188(CONNECTED) 2] ls /configs [myconf] [zk: master:2188(CONNECTED) 3] ls /configs/myconf [solrcore.properties, solrconfig.xml, schema.xml]
3、Tomcat配置与启动
在Tomcat的启动脚本bin/catalina.sh中，增加如下配置：
JAVA_OPTS="-server -Xmx4096m -Xms1024m -verbose:gc -Xloggc:solr_gc.log -Dsolr.solr.home=/home/hadoop/applications/solr/cloud/multicore -DzkHost=master:2188,slave1:2188,slave4:2188"
启动Tomcat服务器：
cd servers/apache-tomcat-7.0.42 bin/catalina.sh start
查看日志：
cd servers/apache-tomcat-7.0.42 tail -100f logs/catalina.out
我们查看一下ZooKeeper中的数据状态，如下所示：
[hadoop@master apache-tomcat-7.0.42]$ cd ~/applications/zookeeper/zookeeper-3.4.5/ [hadoop@master zookeeper-3.4.5]$ bin/zkCli.sh -server master:2188 ... [zk: master:2188(CONNECTED) 0] ls / [configs, zookeeper, clusterstate.json, aliases.json, live_nodes, overseer, overseer_elect, collections] [zk: master:2188(CONNECTED) 1] ls /live_nodes [10.95.3.61:8888_solr-cloud] [zk: master:2188(CONNECTED) 2] ls /collections [collection1]
这时候，SolrCloud集群中只有一个活跃的节点，而且默认生成了一个collection1实例，这个实例实际上虚拟的，因为通过web界面无法访问http://master:8888/solr-cloud/，看不到任何有关SolrCloud的信息，如图所示：
4、同步数据和配置信息，启动其他节点
在另外两个节点上安装Tomcat和Solr服务器，只需要拷贝对应的目录即可：
[hadoop@master ~]$ scp -r servers/ hadoop@slave1:~/ [hadoop@master ~]$ scp -r servers/ hadoop@slave4:~/ [hadoop@master ~]$ scp -r applications/solr/cloud hadoop@slave1:~/applications/solr/ [hadoop@master ~]$ scp -r applications/solr/cloud hadoop@slave4:~/applications/solr/ [hadoop@slave1 ~]$ mkdir -p applications/storage/cloud/data/ [hadoop@slave4 ~]$ mkdir -p applications/storage/cloud/data/
启动其他Solr服务器节点：
[hadoop@slave1 ~]$ cd servers/apache-tomcat-7.0.42 [hadoop@slave1 apache-tomcat-7.0.42]$ bin/catalina.sh start [hadoop@slave4 ~]$ cd servers/apache-tomcat-7.0.42 [hadoop@slave4 apache-tomcat-7.0.42]$ bin/catalina.sh start
查看ZooKeeper集群中数据状态：
[zk: master:2188(CONNECTED) 3] ls /live_nodes [10.95.3.65:8888_solr-cloud, 10.95.3.61:8888_solr-cloud, 10.95.3.62:8888_solr-cloud]
这时，已经存在3个活跃的节点了，但是SolrCloud集群并没有更多信息，访问http://master:8888/solr-cloud/后，同上面的图是一样的，没有SolrCloud相关数据。
5、创建Collection、Shard和Replication
- 创建Collection及初始Shard
直接通过REST接口来创建Collection，如下所示：
[hadoop@master ~]$ curl 'http://master:8888/solr-cloud/admin/collections?action=CREATE&name=mycollection&numShards=3&replicationFactor=1'
如果成功，会输出如下响应内容：
<?xml version="1.0" encoding="UTF-8"?> <response> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">4103</int> </lst> <lst name="success"> <lst> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">3367</int> </lst> <str name="core">mycollection_shard2_replica1</str> <str name="saved">/home/hadoop/applications/solr/cloud/multicore/solr.xml</str> </lst> <lst> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">3280</int> </lst> <str name="core">mycollection_shard1_replica1</str> <str name="saved">/home/hadoop/applications/solr/cloud/multicore/solr.xml</str> </lst> <lst> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">3690</int> </lst> <str name="core">mycollection_shard3_replica1</str> <str name="saved">/home/hadoop/applications/solr/cloud/multicore/solr.xml</str> </lst> </lst> </response>
上面链接中的几个参数的含义，说明如下：
name 待创建Collection的名称 numShards 分片的数量 replicationFactor 复制副本的数量
执行上述操作如果没有异常，已经创建了一个Collection，名称为mycollection，而且每个节点上存在一个分片。这时，也可以查看ZooKeeper中状态：
[zk: master:2188(CONNECTED) 5] ls /collections [mycollection, collection1] [zk: master:2188(CONNECTED) 6] ls /collections/mycollection [leader_elect, leaders]
可以通过Web管理页面，访问http://master:8888/solr-cloud/#/~cloud，查看SolrCloud集群的分片信息，如图所示：
由上图可以看到，对应节点上SOLR分片的对应关系：
shard3 10.95.3.61 master shard1 10.95.3.62 slave1 shard2 10.95.3.65 slave4
实际上，我们从master节点可以看到，SOLR的配置文件内容，已经发生了变化，如下所示：
[hadoop@master ~]$ cat applications/solr/cloud/multicore/solr.xml
<?xml version="1.0" encoding="UTF-8" ?>
<solr persistent="true">
<cores defaultCoreName="collection1" host="${host:}" adminPath="/admin/cores" zkClientTimeout="${zkClientTimeout:15000}" hostPort="8888" hostContext="${hostContext:solr-cloud}">
<core loadOnStartup="true" shard="shard3" instanceDir="mycollection_shard3_replica1/" transient="false" name="mycollection_shard3_replica1" collection="mycollection" />
</cores>
</solr>
- 创建Replication
下面对已经创建的初始分片进行复制。 shard1已经在slave1上，我们复制分片到master和slave4上，执行如下命令：
[hadoop@master ~]$ curl 'http://master:8888/solr-cloud/admin/cores?action=CREATE&collection=mycollection&name=mycollection_shard1_replica_2&shard=shard1' <?xml version="1.0" encoding="UTF-8"?> <response> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">1485</int> </lst> <str name="core">mycollection_shard1_replica_2</str> <str name="saved">/home/hadoop/applications/solr/cloud/multicore/solr.xml</str> </response> [hadoop@master ~]$ curl 'http://master:8888/solr-cloud/admin/cores?action=CREATE&collection=mycollection&name=mycollection_shard1_replica_3&shard=shard1' <?xml version="1.0" encoding="UTF-8"?> <response> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">2543</int> </lst> <str name="core">mycollection_shard1_replica_3</str> <str name="saved">/home/hadoop/applications/solr/cloud/multicore/solr.xml</str> </response> [hadoop@slave4 ~]$ curl 'http://slave4:8888/solr-cloud/admin/cores?action=CREATE&collection=mycollection&name=mycollection_shard1_replica_4&shard=shard1' <?xml version="1.0" encoding="UTF-8"?> <response> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">2405</int> </lst> <str name="core">mycollection_shard1_replica_4</str> <str name="saved">/home/hadoop/applications/solr/cloud/multicore/solr.xml</str> </response>
最后的结果是，slave1上的shard1，在master节点上有2个副本，名称为mycollection_shard1_replica_2和mycollection_shard1_replica_3，在slave4节点上有一个副本，名称为mycollection_shard1_replica_4. 也可以通过查看master和slave4上的目录变化，如下所示：
[hadoop@master ~]$ ll applications/solr/cloud/multicore/ 总用量 24 drwxrwxr-x. 4 hadoop hadoop 4096 8月 1 09:58 collection1 drwxrwxr-x. 3 hadoop hadoop 4096 8月 1 15:41 mycollection_shard1_replica_2 drwxrwxr-x. 3 hadoop hadoop 4096 8月 1 15:42 mycollection_shard1_replica_3 drwxrwxr-x. 3 hadoop hadoop 4096 8月 1 15:23 mycollection_shard3_replica1 -rw-rw-r--. 1 hadoop hadoop 784 8月 1 15:42 solr.xml -rw-rw-r--. 1 hadoop hadoop 1004 8月 1 10:02 zoo.cfg [hadoop@slave4 ~]$ ll applications/solr/cloud/multicore/ 总用量 20 drwxrwxr-x. 4 hadoop hadoop 4096 8月 1 14:53 collection1 drwxrwxr-x. 3 hadoop hadoop 4096 8月 1 15:44 mycollection_shard1_replica_4 drwxrwxr-x. 3 hadoop hadoop 4096 8月 1 15:23 mycollection_shard2_replica1 -rw-rw-r--. 1 hadoop hadoop 610 8月 1 15:44 solr.xml -rw-rw-r--. 1 hadoop hadoop 1004 8月 1 15:08 zoo.cfg
其中，mycollection_shard3_replica1和mycollection_shard2_replica1都是创建Collection的时候自动生成的分片，也就是第一个副本。 通过Web界面，可以更加直观地看到shard1的情况，如图所示：
我们再次从master节点可以看到，SOLR的配置文件内容，又发生了变化，如下所示：
[hadoop@master ~]$ cat applications/solr/cloud/multicore/solr.xml
<?xml version="1.0" encoding="UTF-8" ?>
<solr persistent="true">
<cores defaultCoreName="collection1" host="${host:}" adminPath="/admin/cores" zkClientTimeout="${zkClientTimeout:15000}" hostPort="8888" hostContext="${hostContext:solr-cloud}">
<core loadOnStartup="true" shard="shard3" instanceDir="mycollection_shard3_replica1/" transient="false" name="mycollection_shard3_replica1" collection="mycollection" />
<core loadOnStartup="true" shard="shard1" instanceDir="mycollection_shard1_replica_2/" transient="false" name="mycollection_shard1_replica_2" collection="mycollection" />
<core loadOnStartup="true" shard="shard1" instanceDir="mycollection_shard1_replica_3/" transient="false" name="mycollection_shard1_replica_3" collection="mycollection" />
</cores>
</solr>
到此为止，我们已经基于3个物理节点，配置完成了SolrCloud集群。
索引数据
我们根据前面定义的schema.xml，自己构造了一个数据集，代码如下所示：
package org.shirdrn.solr.data;
import java.io.BufferedWriter;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStreamWriter;
import java.text.DateFormat;
import java.text.SimpleDateFormat;
import java.util.Date;
import java.util.Random;
public class BuildingSampleGenerator {
private final DateFormat df = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss.SSS'Z'");
private Random random = new Random();
static String[] areas = {
"北京", "上海", "深圳", "广州", "天津", "重庆","成都",
"银川", "沈阳", "大连", "吉林", "郑州", "徐州", "兰州",
"东京", "纽约", "贵州", "长春", "大连", "武汉","南京",
"海口", "太原", "济南", "日照", "菏泽", "包头", "松原"
};
long pre = 0L;
long current = 0L;
public synchronized long genId() {
current = System.nanoTime();
if(current == pre) {
try {
Thread.sleep(0, 1);
} catch (InterruptedException e) {
e.printStackTrace();
}
current = System.nanoTime();
pre = current;
}
return current;
}
public String genArea() {
return areas[random.nextInt(areas.length)];
}
private int maxLatitude = 90;
private int maxLongitude = 180;
public Coordinate genCoordinate() {
int beforeDot = random.nextInt(maxLatitude);
double afterDot = random.nextDouble();
double lat = beforeDot + afterDot;
beforeDot = random.nextInt(maxLongitude);
afterDot = random.nextDouble();
double lon = beforeDot + afterDot;
return new Coordinate(lat, lon);
}
private Random random1 = new Random(System.currentTimeMillis());
private Random random2 = new Random(2 * System.currentTimeMillis());
public int genFloors() {
return 1 + random1.nextInt(50) + random2.nextInt(50);
}
public class Coordinate {
double latitude;
double longitude;
public Coordinate() {
super();
}
public Coordinate(double latitude, double longitude) {
super();
this.latitude = latitude;
this.longitude = longitude;
}
public double getLatitude() {
return latitude;
}
public double getLongitude() {
return longitude;
}
}
static int[] signs = {-1, 1};
public int genTemperature() {
return signs[random.nextInt(2)] * random.nextInt(81);
}
static String[] codes = {"A", "B", "C", "D", "E", "F", "G", "H", "I",
"J", "K", "L", "M", "N", "O", "P", "Q", "R", "S", "T", "U", "V",
"W", "X", "Y", "Z"};
public String genCode() {
return codes[random.nextInt(codes.length)];
}
static int[] types = {0, 1, 2, 3};
public int genBuildingType() {
return types[random.nextInt(types.length)];
}
static String[] categories = {
"办公建筑", "教育建筑", "商业建筑", "文教建筑", "医卫建筑",
"住宅", "宿舍", "公寓", "工业建筑"};
public String genBuildingCategory() {
return categories[random.nextInt(categories.length)];
}
public void generate(String file, int count) throws IOException {
BufferedWriter w = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(file), "UTF-8"));
w.write("id,area,building_type,category,temperature,code,latitude,longitude,when");
w.newLine();
for(int i=0; i<count; i++) {
String when = df.format(new Date());
StringBuffer sb = new StringBuffer();
sb.append(genId()).append(",")
.append("\"").append(genArea()).append("\"").append(",")
.append(genBuildingType()).append(",")
.append("\"").append(genBuildingCategory()).append("\"").append(",")
.append(genTemperature()).append(",")
.append(genCode()).append(",");
Coordinate coord = genCoordinate();
sb.append(coord.latitude).append(",")
.append(coord.longitude).append(",")
.append("\"").append(when).append("\"");
w.write(sb.toString());
w.newLine();
}
w.close();
System.out.println("Finished: file=" + file);
}
public static void main(String[] args) throws Exception {
BuildingSampleGenerator gen = new BuildingSampleGenerator();
String file = "E:\\Develop\\eclipse-jee-kepler\\workspace\\solr-data\\building_files";
for(int i=0; i<=9; i++) {
String f = new String(file + "_100w_0" + i + ".csv");
gen.generate(f, 5000000);
}
}
}
生成的文件，如下所示：
[hadoop@master solr-data]$ ll building_files_100w* -rw-rw-r--. 1 hadoop hadoop 109025853 7月 26 14:05 building_files_100w_00.csv -rw-rw-r--. 1 hadoop hadoop 108015504 7月 26 10:53 building_files_100w_01.csv -rw-rw-r--. 1 hadoop hadoop 108022184 7月 26 11:00 building_files_100w_02.csv -rw-rw-r--. 1 hadoop hadoop 108016854 7月 26 11:00 building_files_100w_03.csv -rw-rw-r--. 1 hadoop hadoop 108021750 7月 26 11:00 building_files_100w_04.csv -rw-rw-r--. 1 hadoop hadoop 108017496 7月 26 11:00 building_files_100w_05.csv -rw-rw-r--. 1 hadoop hadoop 108016193 7月 26 11:00 building_files_100w_06.csv -rw-rw-r--. 1 hadoop hadoop 108023537 7月 26 11:00 building_files_100w_07.csv -rw-rw-r--. 1 hadoop hadoop 108014684 7月 26 11:00 building_files_100w_08.csv -rw-rw-r--. 1 hadoop hadoop 108022044 7月 26 11:00 building_files_100w_09.csv
数据文件格式如下：
[hadoop@master solr-data]$ head building_files_100w_00.csv id,area,building_type,category,temperature,code,latitude,longitude,when 18332617097417,"广州",2,"医卫建筑",61,N,5.160762478343409,62.92919119315037,"2013-07-26T14:05:55.832Z" 18332617752331,"成都",1,"教育建筑",10,Q,77.34792453477195,72.59812030045762,"2013-07-26T14:05:55.833Z" 18332617815833,"大连",0,"教育建筑",18,T,81.47569061530493,0.2177194388096203,"2013-07-26T14:05:55.833Z" 18332617903711,"广州",0,"办公建筑",31,D,51.85825084513671,13.60710950097155,"2013-07-26T14:05:55.833Z" 18332617958555,"深圳",3,"商业建筑",5,H,22.181374031472675,119.76001810254823,"2013-07-26T14:05:55.833Z" 18332618020454,"济南",3,"公寓",-65,L,84.49607030736806,29.93095171443135,"2013-07-26T14:05:55.834Z" 18332618075939,"北京",2,"住宅",-29,J,86.61660177436184,39.20847527640485,"2013-07-26T14:05:55.834Z" 18332618130141,"菏泽",0,"医卫建筑",24,J,70.57574551258345,121.21977908377244,"2013-07-26T14:05:55.834Z" 18332618184343,"徐州",2,"办公建筑",31,W,0.10129771041097524,153.40533210345387,"2013-07-26T14:05:55.834Z"
我们向已经搭建好的SolrCloud集群，执行索引数据的操作。这里，实现了一个简易的客户端，代码如下所示：
package org.shirdrn.solr.indexing;
import java.io.IOException;
import java.net.MalformedURLException;
import java.text.DateFormat;
import java.text.SimpleDateFormat;
import java.util.Date;
import org.apache.solr.client.solrj.SolrServerException;
import org.apache.solr.client.solrj.impl.CloudSolrServer;
import org.apache.solr.common.SolrInputDocument;
import org.shirdrn.solr.data.BuildingSampleGenerator;
import org.shirdrn.solr.data.BuildingSampleGenerator.Coordinate;
public class CloudSolrClient {
private CloudSolrServer cloudSolrServer;
public synchronized void open(final String zkHost, final String defaultCollection,
int zkClientTimeout, final int zkConnectTimeout) {
if (cloudSolrServer == null) {
try {
cloudSolrServer = new CloudSolrServer(zkHost);
cloudSolrServer.setDefaultCollection(defaultCollection);
cloudSolrServer.setZkClientTimeout(zkClientTimeout);
cloudSolrServer.setZkConnectTimeout(zkConnectTimeout);
} catch (MalformedURLException e) {
System.out
.println("The URL of zkHost is not correct!! Its form must as below:\n zkHost:port");
e.printStackTrace();
} catch (Exception e) {
e.printStackTrace();
}
}
}
public void addDoc(long id, String area, int buildingType, String category,
int temperature, String code, double latitude, double longitude, String when) {
try {
SolrInputDocument doc = new SolrInputDocument();
doc.addField("id", id);
doc.addField("area", area);
doc.addField("building_type", buildingType);
doc.addField("category", category);
doc.addField("temperature", temperature);
doc.addField("code", code);
doc.addField("latitude", latitude);
doc.addField("longitude", longitude);
doc.addField("when", when);
cloudSolrServer.add(doc);
cloudSolrServer.commit();
} catch (SolrServerException e) {
System.err.println("Add docs Exception !!!");
e.printStackTrace();
} catch (IOException e) {
e.printStackTrace();
} catch (Exception e) {
System.err.println("Unknowned Exception!!!!!");
e.printStackTrace();
}
}
public static void main(String[] args) {
final String zkHost = "master:2188";
final String defaultCollection = "mycollection";
final int zkClientTimeout = 20000;
final int zkConnectTimeout = 1000;
CloudSolrClient client = new CloudSolrClient();
client.open(zkHost, defaultCollection, zkClientTimeout, zkConnectTimeout);
BuildingSampleGenerator gen = new BuildingSampleGenerator();
final DateFormat df = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss.SSS'Z'");
for(int i = 0; i < 10000; i++) {
long id = gen.genId();
String area = gen.genArea();
int buildingType = gen.genBuildingType();
String category = gen.genBuildingCategory();
int temperature = gen.genTemperature();
String code = gen.genCode();
Coordinate coord = gen.genCoordinate();
double latitude = coord.getLatitude();
double longitude = coord.getLongitude();
String when = df.format(new Date());
client.addDoc(id, area, buildingType, category, temperature, code, latitude, longitude, when);
}
}
}
这样，可以查看SolrCloud管理页面，或者直接登录到服务器上，能够看到对应索引数据分片的情况，比较均匀地分布到各个Shard节点上。 当然，也可以从Web管理页面上来管理各个分片的副本数据，比如某个分片具有太多的副本，通过页面上的删除掉（unload）该副本，实际该副本的元数据信息被从ZooKeeper集群维护的信息中删除，在具体的节点上的副本数据并没有删除，而只是处于离线状态，不能提供服务。
搜索数据
我们可以执行搜索，执行如下搜索条件：
http://master:8888/solr-cloud/mycollection/select?q=北京 纽约&fl=*&fq=category:公寓&fq=building_type:2&start=0&rows=10
搜索结果，如下所示：
<response> <lst name="responseHeader"> <int name="status">0</int> <int name="QTime">570</int> </lst> <result name="response" numFound="201568" start="0" maxScore="1.5322487"> <doc> <long name="id">37109751480918</long> <long name="_version_">1442164237143113728</long> </doc> <doc> <long name="id">37126929150371</long> <long name="_version_">1442164255154503680</long> </doc> <doc> <long name="id">37445266827945</long> <long name="_version_">1442164588949798912</long> </doc> <doc> <long name="id">37611390043867</long> <long name="_version_">1442164763138195456</long> </doc> <doc> <long name="id">37892268870281</long> <long name="_version_">1442165057653833728</long> </doc> <doc> <long name="id">89820941817153</long> <long name="_version_">1442219517734289408</long> </doc> <doc> <long name="id">89825667635450</long> <long name="_version_">1442219522665742336</long> </doc> <doc> <long name="id">89830029550692</long> <long name="_version_">1442219527207124993</long> </doc> <doc> <long name="id">93932235463589</long> <long name="_version_">1442223828610580480</long> </doc> <doc> <long name="id">93938975733467</long> <long name="_version_">1442223835684274177</long> </doc> </result> </response>
可以查看对应的日志，示例如下所示：
2013-08-05 18:38:26.814 [http-bio-8888-exec-228] INFO org.apache.solr.core.SolrCore – [mycollection_shard1_0_replica2] webapp=/solr-cloud path=/select params={NOW=1375699145633&shard.url=10.95.3.62:8888/solr-cloud/mycollection_shard1_0_replica1/|10.95.3.61:8888/solr-cloud/mycollection_shard1_0_replica3/&fl=id,score&start=0&q=北京+纽约&distrib=false&wt=javabin&isShard=true&fsv=true&fq=category:公寓&fq=building_type:2&version=2&rows=10} hits=41529 status=0 QTime=102
2013-08-05 18:39:06.203 [http-bio-8888-exec-507] INFO org.apache.solr.core.SolrCore – [mycollection_shard3_replica1] webapp=/solr-cloud path=/select params={fl=*&start=0&q=北京+纽约&fq=category:公寓&fq=building_type:2&rows=10} hits=201568 status=0 QTime=570
相关问题
1、我在进行Collection的创建的时候，当前有4个节点，在ZooKeeper集群中注册，执行如下命令：
[hadoop@slave1 multicore]$ curl 'http://slave1:8888/solr-cloud/admin/collections?action=CREATE&name=tinycollection&numShards=2&replicationFactor=3'
出现异常：
<?xml version="1.0" encoding="UTF-8"?> <response> <lst name="responseHeader"> <int name="status">400</int> <int name="QTime">81</int> </lst> <str name="Operation createcollection caused exception:">org.apache.solr.common.SolrException:org.apache.solr.common.SolrException: Cannot create collection tinycollection. Value of maxShardsPerNode is 1, and the number of live nodes is 4. This allows a maximum of 4 to be created. Value of numShards is 2 and value of replicationFactor is 3. This requires 6 shards to be created (higher than the allowed number)</str> <lst name="exception"> <str name="msg">Cannot create collection tinycollection. Value of maxShardsPerNode is 1, and the number of live nodes is 4. This allows a maximum of 4 to be created. Value of numShards is 2 and value of replicationFactor is 3. This requires 6 shards to be created (higher than the allowed number)</str> <int name="rspCode">400</int> </lst> <lst name="error"> <str name="msg">Cannot create collection tinycollection. Value of maxShardsPerNode is 1, and the number of live nodes is 4. This allows a maximum of 4 to be created. Value of numShards is 2 and value of replicationFactor is 3. This requires 6 shards to be created (higher than the allowed number)</str> <int name="code">400</int> </lst> </response>
根据上面异常信息可知，当前有4个节点可用，但是我在创建Collection的时候，指定两个Shard，同时复制因子是3，所以最低要求，需要6个节点。所以，可以减少复制因子，例如replicationFactor=2，表示一共存在两个副本（Leader分片和另一个副本），然后再执行创建Collection的操作就不会报错了。
本文基于署名-非商业性使用-相同方式共享 4.0许可协议发布，欢迎转载、使用、重新发布，但务必保留文章署名时延军（包含链接：http://shiyanjun.cn），不得用于商业目的，基于本文修改后的作品务必以相同的许可发布。如有任何疑问，请与我联系。
写的真好