download zookeeper
make a copy of conf\zk_sample.cfg as zoo.cfg
creaeta zkdata folder.
in zoo.cfg, uncomment datadir and provide path to zkdata folder (replace '\' in the path with '/' )
Modify zk_home/bin/zk_env.cmd to replace ~dp0.. with actual paths like below
REM assuming Zookeeper is installed at : D:\zk1
set ZOOCFGDIR=D:\zk1\conf
set ZOO_LOG_DIR=D:\zk1
set ZOO_LOG4J_PROP=INFO,CONSOLE
set CLASSPATH=%ZOOCFGDIR%
SET CLASSPATH=D:\zk1\*;D:\zk1\lib\*;%CLASSPATH%
SET CLASSPATH=D:\zk1\build\classes;D:\zk1\build\lib\*;%CLASSPATH%
set ZOOCFG=%ZOOCFGDIR%\zoo.cfg
Lets assume you have a stand alone solr with "collection1" core and we want to convert this into soldCloud cluster and split this core into multiple shards on diff machine.
1) First start zookeepr, run following command from zk_home/bin path:
zkServer.cmd
2) Next upload & link your standalone core config to Zookeeper (Assuming Solr is installed at : D:\solr-4.4.0\mySolr1)
make sure to use '/' as path seprate instead of '\' for the confdir path below
cd D:\solr-4.4.0\mySolr1\cloud-scripts
zkcli.bat -zkhost localhost:2181 -cmd upconfig -confdir D:/solr-4.4.0/mySolr1/solr/collection1/conf -confname collection1cfg
zkcli.bat -zkhost localhost:2181 -cmd linkconfig -collection collection1 -confname collection1cfg
3) Next start your solr with following command (assuming zookeeper is running on localhost:2181) and solr is installed at D:\solr-4.4.0\mySolr1
java -Xmx2g -Djetty.port=8983 -DzkHost=localhost:2181 -Dsolr.solr.home=D:\solr-4.4.0\mySolr1\solr -jar start.jar
4) Go to localhost:8989/solr and verify the # of documents in your core with following URL:
http://localhost:8983/solr/collection1/select?q=*:*&rows=0
make a note of numFound value in the response returned.
Also verify that following URL shows the "collection1" core we have as "shard1"
http://localhost:8983/solr/#/~cloud
5) Next in the browser run the following command to split your core "collection1" into multiple shards (assuming shard1 is the name your shard and collection1 is the name of your core)
http://localhost:8983/solr/admin/collections?action=SPLITSHARD&collection=collection1&shard=shard1&async=5
This should generated 2 shards / cores
collection1_shard1_1_replica1
collection1_shard1_0_replica1
6) Next on another machine (e.g. machine2) , start solr with following command (assuming zookeeper is running on <>:2181 (replace "machine1" with name or ip address of machine1)
and solr on machine 2 is installed at : D:/solr-4.4.0/mySolr2
java -Xmx2g -Djetty.port=7574 -DzkHost=machine1:2181 -Dsolr.solr.home=D:/solr-4.4.0/mySolr2/solr -jar start.jar
This will start solr on port 7574 on machine 2
7) On machine2, Run following command to create a new Core (collection1_shard1_1_core2) specifying the collection=collection1 (same as original core name on machine1 before splitting into shards)
and shard=shard1_1 (same as what shown on http://localhost:8983/solr/#/~cloud page for the shard that you want to move to machine2)
http://localhost:7574/solr/admin/cores?action=CREATE&collection=collection1&shard=shard1_1&name=collection1_shard1_1_core2
8) now shutdown solr (not zookeeper) running on machine1 by pressing ctrl + C in the running solr window on machine 1.
Move the shard1_1_replica1 folder from solr installation, to some other location outside SOLR_HOME\solr directory.
And re-run the solr on machine 1
java -Xmx2g -Djetty.port=8983 -DzkHost=localhost:2181 -Dsolr.solr.home=D:\solr-4.4.0\mySolr1\solr -jar start.jar
Verify now shard on machine 2 is shown as shard leader for shard1_1.
We have now sucessfully converted a single core named 'collection1' into a distributed collection 'collection1' with shard 'shard1_0' on original machine1 and moved shard1_1 to another machine2.
make a copy of conf\zk_sample.cfg as zoo.cfg
creaeta zkdata folder.
in zoo.cfg, uncomment datadir and provide path to zkdata folder (replace '\' in the path with '/' )
Modify zk_home/bin/zk_env.cmd to replace ~dp0.. with actual paths like below
REM assuming Zookeeper is installed at : D:\zk1
set ZOOCFGDIR=D:\zk1\conf
set ZOO_LOG_DIR=D:\zk1
set ZOO_LOG4J_PROP=INFO,CONSOLE
set CLASSPATH=%ZOOCFGDIR%
SET CLASSPATH=D:\zk1\*;D:\zk1\lib\*;%CLASSPATH%
SET CLASSPATH=D:\zk1\build\classes;D:\zk1\build\lib\*;%CLASSPATH%
set ZOOCFG=%ZOOCFGDIR%\zoo.cfg
Lets assume you have a stand alone solr with "collection1" core and we want to convert this into soldCloud cluster and split this core into multiple shards on diff machine.
1) First start zookeepr, run following command from zk_home/bin path:
zkServer.cmd
2) Next upload & link your standalone core config to Zookeeper (Assuming Solr is installed at : D:\solr-4.4.0\mySolr1)
make sure to use '/' as path seprate instead of '\' for the confdir path below
cd D:\solr-4.4.0\mySolr1\cloud-scripts
zkcli.bat -zkhost localhost:2181 -cmd upconfig -confdir D:/solr-4.4.0/mySolr1/solr/collection1/conf -confname collection1cfg
zkcli.bat -zkhost localhost:2181 -cmd linkconfig -collection collection1 -confname collection1cfg
3) Next start your solr with following command (assuming zookeeper is running on localhost:2181) and solr is installed at D:\solr-4.4.0\mySolr1
java -Xmx2g -Djetty.port=8983 -DzkHost=localhost:2181 -Dsolr.solr.home=D:\solr-4.4.0\mySolr1\solr -jar start.jar
4) Go to localhost:8989/solr and verify the # of documents in your core with following URL:
http://localhost:8983/solr/collection1/select?q=*:*&rows=0
make a note of numFound value in the response returned.
Also verify that following URL shows the "collection1" core we have as "shard1"
http://localhost:8983/solr/#/~cloud
5) Next in the browser run the following command to split your core "collection1" into multiple shards (assuming shard1 is the name your shard and collection1 is the name of your core)
http://localhost:8983/solr/admin/collections?action=SPLITSHARD&collection=collection1&shard=shard1&async=5
This should generated 2 shards / cores
collection1_shard1_1_replica1
collection1_shard1_0_replica1
6) Next on another machine (e.g. machine2) , start solr with following command (assuming zookeeper is running on <
and solr on machine 2 is installed at : D:/solr-4.4.0/mySolr2
java -Xmx2g -Djetty.port=7574 -DzkHost=machine1:2181 -Dsolr.solr.home=D:/solr-4.4.0/mySolr2/solr -jar start.jar
This will start solr on port 7574 on machine 2
7) On machine2, Run following command to create a new Core (collection1_shard1_1_core2) specifying the collection=collection1 (same as original core name on machine1 before splitting into shards)
and shard=shard1_1 (same as what shown on http://localhost:8983/solr/#/~cloud page for the shard that you want to move to machine2)
http://localhost:7574/solr/admin/cores?action=CREATE&collection=collection1&shard=shard1_1&name=collection1_shard1_1_core2
8) now shutdown solr (not zookeeper) running on machine1 by pressing ctrl + C in the running solr window on machine 1.
Move the shard1_1_replica1 folder from solr installation, to some other location outside SOLR_HOME\solr directory.
And re-run the solr on machine 1
java -Xmx2g -Djetty.port=8983 -DzkHost=localhost:2181 -Dsolr.solr.home=D:\solr-4.4.0\mySolr1\solr -jar start.jar
Verify now shard on machine 2 is shown as shard leader for shard1_1.
We have now sucessfully converted a single core named 'collection1' into a distributed collection 'collection1' with shard 'shard1_0' on original machine1 and moved shard1_1 to another machine2.
