This all works well. However if we end up with more inputs defined each one starts it's own Java process. They don't use much CPU (~0.4%) or memory (~150 MB) but in the interests of keeping things efficient I was hopeful the host for the data could be pulled from the jmxserver definition even if it was a separate attribute.
We have tested the combined config.xml as you can have multiple jmxserver definitions in it and it works fine. However everything gets marked with the same host value.
I am aware I could probably put a props.conf and transforms.conf to extract it from the raw event and convert the host at index time but again that adds to the complexity of the entire configuration for anyone in the future who might want to add more jmxserver configurations.
I'm trying to keep the whole process as easy as possible so that future users who are less Splunk savvy will be able to quickly and easily add more jmx servers but just editting the one xml config file.
So basically after all of that is it possible to define the host that the input will use in the jmxserver definition allowing for less java processes running on the collection endpoint?