metricbeat: add more node metrics in elasticseach module

[wangdisdu]

metricbeat: add more node metrics in elastics each module

node:

• jvm.memory.heap.init
• jvm.memory.heap.max
• jvm.memory.nonheap.init
• jvm.memory.nonheap.max
• process.mlockall

[elasticmachine]

Can one of the admins verify this patch?

[ruflin]

@wangdisdu Could you share some more details on how you use these metrics in your environment?

[wangdisdu]

@ruflin Well, i build many charts with echart in use of "node stats" metrics. With the charts showing jvm memory used by heap and nonheap, i am able to add two base lines: init line, and max line, so that for example, i can figure out unhealthy state when the used line is approaching the max line.

For the process.mlockall, it is very important for me. if it is false, means system environment configuration is incorrectly set, i need to fix it and restart elasticsearch.

[ruflin]

@wangdisdu Thanks for sharing the details. How does the data from beats end up in echarts? Is that on top of elasticsearch?

We are looking into which values we should add to elasticsearch module without adding too many values. We will keep you posted.

[ruflin]

Note for myself: this is a breaking change.

[ruflin]

@wangdisdu I think we should move forward with this PR. Could you add a line to the CHANGELOG under breaking changes as the heap.init which was heap_init is a breaking change?

[wangdisdu]

@ruflin Thanks for your reviewed. I has added the breaking change log into CHANGELOG: Rename heap_init field to heap.init.

By the way, the echarts is just a javascript libaray. The echarts can be used to build beautiful chart with the metric data that beats collect and store into elasticsearch.

[ruflin]

jenkins, test it

[ruflin]

@wangdisdu Thanks for your contribution.