Format of solr.xml

The solr.xml file defines some global configuration options that apply to all or many cores.

This section will describe the default solr.xml file included with Solr and how to modify it for your needs. For details on how to configure core.properties, see the section Defining core.properties.

Defining solr.xml

You can find solr.xml in your $SOLR_HOME directory (usually server/solr) in standalone mode or in ZooKeeper when using SolrCloud. The default solr.xml file looks like this:

<solr>

  <solrcloud>
    <str name="host">${host:}</str>
    <int name="hostPort">${jetty.port:8983}</int>
    <str name="hostContext">${hostContext:solr}</str>
    <int name="zkClientTimeout">${zkClientTimeout:15000}</int>
    <bool name="genericCoreNodeNames">${genericCoreNodeNames:true}</bool>
  </solrcloud>

  <shardHandlerFactory name="shardHandlerFactory"
    class="HttpShardHandlerFactory">
    <int name="socketTimeout">${socketTimeout:0}</int>
    <int name="connTimeout">${connTimeout:0}</int>
  </shardHandlerFactory>

</solr>

As you can see, the discovery Solr configuration is "SolrCloud friendly". However, the presence of the <solrcloud> element does not mean that the Solr instance is running in SolrCloud mode. Unless the -DzkHost or -DzkRun are specified at startup time, this section is ignored.

Solr.xml Parameters

The <solr> Element

There are no attributes that you can specify in the <solr> tag, which is the root element of solr.xml. The tables below list the child nodes of each XML element in solr.xml.

Node Description

adminHandler

If used, this attribute should be set to the FQN (Fully qualified name) of a class that inherits from CoreAdminHandler. For example, <str name="adminHandler">com.myorg.MyAdminHandler</str> would configure the custom admin handler (MyAdminHandler) to handle admin requests. If this attribute isn’t set, Solr uses the default admin handler, org.apache.solr.handler.admin.CoreAdminHandler. For more information on this parameter, see the Solr Wiki at http://wiki.apache.org/solr/CoreAdmin#cores.

collectionsHandler

As above, for custom CollectionsHandler implementations.

infoHandler

As above, for custom InfoHandler implementations.

coreLoadThreads

Specifies the number of threads that will be assigned to load cores in parallel.

coreRootDirectory

The root of the core discovery tree, defaults to $SOLR_HOME.

managementPath

Currently non-operational.

sharedLib

Specifies the path to a common library directory that will be shared across all cores. Any JAR files in this directory will be added to the search path for Solr plugins. This path is relative to $SOLR_HOME. Custom handlers may be placed in this directory.

shareSchema

This attribute, when set to true, ensures that the multiple cores pointing to the same Schema resource file will be referring to the same IndexSchema Object. Sharing the IndexSchema Object makes loading the core faster. If you use this feature, make sure that no core-specific property is used in your Schema file.

transientCacheSize

Defines how many cores with transient=true that can be loaded before swapping the least recently used core for a new core.

configSetBaseDir

The directory under which configsets for Solr cores can be found. Defaults to $SOLR_HOME/configsets.

The <solrcloud> Element

This element defines several parameters that relate so SolrCloud. This section is ignored unless theSolr instance is started with either -DzkRun or -DzkHost

Node Description

distribUpdateConnTimeout

Used to set the underlying "connTimeout" for intra-cluster updates.

distribUpdateSoTimeout

Used to set the underlying "socketTimeout" for intra-cluster updates.

host

The hostname Solr uses to access cores.

hostContext

The url context path.

hostPort

The port Solr uses to access cores. In the default solr.xml file, this is set to ${jetty.port:8983}, which will use the Solr port defined in Jetty, and otherwise fall back to 8983.

leaderVoteWait

When SolrCloud is starting up, how long each Solr node will wait for all known replicas for that shard to be found before assuming that any nodes that haven’t reported are down.

leaderConflictResolveWait

When trying to elect a leader for a shard, this property sets the maximum time a replica will wait to see conflicting state information to be resolved; temporary conflicts in state information can occur when doing rolling restarts, especially when the node hosting the Overseer is restarted. Typically, the default value of 180000 (ms) is sufficient for conflicts to be resolved; you may need to increase this value if you have hundreds or thousands of small collections in SolrCloud.

zkClientTimeout

A timeout for connection to a ZooKeeper server. It is used with SolrCloud.

zkHost

In SolrCloud mode, the URL of the ZooKeeper host that Solr should use for cluster state information.

genericCoreNodeNames

If TRUE, node names are not based on the address of the node, but on a generic name that identifies the core. When a different machine takes over serving that core things will be much easier to understand.

zkCredentialsProvider & ` zkACLProvider`

Optional parameters that can be specified if you are using ZooKeeper Access Control.

The <logging> Element

Node Description

class

The class to use for logging. The corresponding JAR file must be available to Solr, perhaps through a <lib> directive in solrconfig.xml.

enabled

true/false - whether to enable logging or not.

The <logging><watcher> Element
Node Description

size

The number of log events that are buffered.

threshold

The logging level above which your particular logging implementation will record. For example when using log4j one might specify DEBUG, WARN, INFO, etc.

The <shardHandlerFactory> Element

Custom shard handlers can be defined in solr.xml if you wish to create a custom shard handler.

<shardHandlerFactory name="ShardHandlerFactory" class="qualified.class.name">

Since this is a custom shard handler, sub-elements are specific to the implementation. The default and only shard handler provided by Solr is the HttpShardHandlerFactory in which case, the following sub-elements can be specified:

Node Description

socketTimeout

The read timeout for intra-cluster query and administrative requests. The default is the same as the distribUpdateSoTimeout specified in the <solrcloud> section.

connTimeout

The connection timeout for intra-cluster query and administrative requests. Defaults to the distribUpdateConnTimeout specified in the <solrcloud> section

urlScheme

URL scheme to be used in distributed search

maxConnectionsPerHost

Maximum connections allowed per host. Defaults to 20

maxConnections

Maximum total connections allowed. Defaults to 10000

corePoolSize

The initial core size of the threadpool servicing requests. Default is 0.

maximumPoolSize

The maximum size of the threadpool servicing requests. Default is unlimited.

maxThreadIdleTime

The amount of time in seconds that idle threads persist for in the queue, before being killed. Default is 5 seconds.

sizeOfQueue

If the threadpool uses a backing queue, what is its maximum size to use direct handoff. Default is to use a SynchronousQueue.

fairnessPolicy

A boolean to configure if the threadpool favours fairness over throughput. Default is false to favor throughput.

Substituting JVM System Properties in solr.xml

Solr supports variable substitution of JVM system property values in solr.xml, which allows runtime specification of various configuration options. The syntax is ${propertyname[:option default value]}. This allows defining a default that can be overridden when Solr is launched. If a default value is not specified, then the property must be specified at runtime or the solr.xml file will generate an error when parsed.

Any JVM system properties usually specified using the -D flag when starting the JVM, can be used as variables in the solr.xml file.

For example, in the solr.xml file shown below, the socketTimeout and connTimeout values are each set to "0". However, if you start Solr using ‘bin/solr -DsocketTimeout=1000’, the socketTimeout option of the HttpShardHandlerFactory to be overridden using a value of 1000ms, while the connTimeout option will continue to use the default property value of "0".

<solr>
  <shardHandlerFactory name="shardHandlerFactory"
                       class="HttpShardHandlerFactory">
    <int name="socketTimeout">${socketTimeout:0}</int>
    <int name="connTimeout">${connTimeout:0}</int>
  </shardHandlerFactory>
</solr>