Does hunk-mongoDB spawn a separate process per inline search. Also I would want to know if I have a
Dashboard with multiple panels, does the "App for mongoDB" leverage the single jvm processes for a dashboard?
The JVM behavior is in your control.
When Hunk calls MongoDB it requires the MongoDB provider. The provider starts a Java Process to connect and brings the results back to Splunk. The connection to MongoDB is control by the below configuration - you can modify this behavior.
vix.command = java
vix.command.arg.1 = -Xmx512m
vix.command.arg.2 = -classpath
vix.command.arg.3 = $SPLUNK_HOME/bin/jars/SplunkMR-s6.0-h1.0.jar:$SPLUNK_HOME/etc/apps/MongoDBApp/bin/hunk_mongodb_app.jar:$SPLUNK_HOME/etc/apps/MongoDBApp/bin/lib/*
vix.command.arg.4 = com.splunk.erp.mongodb.MongoDBERP
To be more clear, If 5-10 query started in parallel, would it use the same JVM with 5-10 threads or it would spawn a separate JVM for each query?
Exactly. After the provider opens a java connection to hadoop, we keep the JVM running until we get all the results or a heartbeat timeout kills the connection. In the case of a provider with MongoDB, we close the connection after the stream from MongoDB gets all the results.
Thanks, radagan.
That means Hunk will start a separate provider (JVM instance), using the provider settings, for each Splunk query?