Metrics about the scrape itself

[finkr]

To monitor the exporter effectively, it would be nice to expose some internals as metrics. This introspection is explained in Prometheus documentation (Writing exporters)

Currently those metrics are expose in log only, so it's difficult to graph and alerts those data.

vcenter testserver: setting up query inventory
vcenter testserver: connecting
vcenter testserver: skipped 106 objects because they are either not connected or not powered on
vcenter testserver: 24 objects (21 vm and 3 hosts)
vcenter testserver: queries generated
vcenter testserver: 24 queries
vcenter testserver: 450 total metricIds
vcenter testserver: 1350 total counter (accounting for 3 instances ratio)
vcenter testserver: created batch 0 from queries 1 - 24
vcenter testserver: 1 threads generated to execute queries
vcenter testserver: thread 1 requests 24 metrics
vcenter testserver thread 1: requesting 24 metrics
vcenter testserver thread 1: retuned 24 metrics
vcenter testserver thread 1: 24 objects (21 vm and 3 hosts)
vcenter testserver: disconnecting
memory usage: sys=15.3M alloc=2.5M

Note: Prometheus automatically create some metrics like : scrape_samples_scraped, and Prometheus library make it easy to expose the exporter consumption (see https://github.com/cblomart/vsphere-graphite/issues/98 )