Service metrics: memory, processor and request volume
Which numbers actually tell you whether a service is healthy, how to read a memory graph before a crash, and where to see your usage on Voroa right now.
What this covers
Four numbers explain most of what happens to a small service. Memory tells you whether you are heading for a crash, and a line that climbs steadily and never comes down is a leak that will eventually kill the process. Processor use tells you whether requests are getting slower because your own code is busy. Request volume gives everything else context, since a rise in both memory and processor is only worrying if traffic did not rise with it. Response time is what your users actually feel, and is the one to watch after every deploy.
Detailed metric graphs and history are not available on Voroa today. Current memory and processor use for each service are shown in the dashboard. Processor use is shown as an average rather than a reading from one instant, so a brief spike may not appear there while sustained heavy use will show up in the average.