Reputation: 7262
I'm upgrading ElasticSearch from 1.2 to 2.2. I dug successfully through quite some breaking API changes. I found that groovy scripts in script_score don't work. I enabled dynamic scripting by turning on
script.inline=true
script.indexed=true
I included groovy-all 2.4.6. I also added jna and mustache to not see any exceptions when the embedded ElasticSearch starts.
The configuration of my embedded server is:
ESLoggerFactory.setDefaultFactory(new Slf4jESLoggerFactory());
Settings settings = Settings.builder()
.put("node.name" ,getName())
.put("path.home", "/tmp/elastic-search/home")
.put("path.shared_data", /tmp/elastic-search")
.put(IndexMetaData.SETTING_NUMBER_OF_SHARDS, "1")
.put(IndexMetaData.SETTING_NUMBER_OF_REPLICAS, "0")
.put("action.auto_create_index", "0")
.put("index.gateway.type", "none")
.put("script.inline", true)
.put("script.indexed", true)
.put("action.destructive_requires_name", false)
.build();
node = nodeBuilder().clusterName(clusterName).settings(settings).node();
Then I execute a query which contains a snippet under function_score/functions which looks loke:
"filter" : {
"exists" : {
"field" : "transactionCount"
}
},
"script_score" : {
"script" : {
"inline" : "doc['transactionCount'].value/10.0"
}
}
The execution throws:
Caused by: org.elasticsearch.index.query.QueryParsingException: script_score the script could not be loaded
...
Caused by: java.lang.IllegalArgumentException: script_lang not supported [groovy]
at org.elasticsearch.script.ScriptService.getScriptEngineServiceForLang(ScriptService.java:211)
When debugging I saw in ScriptModule that the script engine for Groovy is missing. Only the native and the mustache ones are present.
Any tip is very welcome:)
Upvotes: 3
Views: 6763
Reputation: 31
Add the below 3 entries in the elasticsearch.yml file. It will resolve the issue.
index.max_result_window: 2147483647
script.engine.groovy.inline.aggs: on
script.engine.groovy.inline.search: on
Upvotes: 0
Reputation: 11022
In ElasticSearch 2.2, the scripts engines have been externalized: they are now plugins, which you can install on demand. Plugins are discovered by scanning the plugins
folder.
If you use an embedded node, you have two choices:
In my project, I use the 2nd case: ElasticSearch is embedded in an osgi container, and I wanted to discover the plugins through a classical classpath scanning. To do this kind of things, you must override Node
:
class ConfigurableNode extends Node {
public ConfigurableNode(Settings settings, Collection<Class<? extends Plugin>> classpathPlugins) {
super(InternalSettingsPreparer.prepareEnvironment(preparedSettings, null),
Version.CURRENT,
classpathPlugins);
}
}
and you create your node with :
Node node = new ConfigurableNode(nodeBuilder()
.clusterName(clusterName)
.settings(settings)
.getSettings(),
Arrays.asList(GroovyPlugin.class));
It's not pretty but I didn't find a method more elegant in v2.2
Upvotes: 8