Runner health metrics

[erikh]

We need to add these to the runner and queuesvc/datasvc/uisvc or remove them from the initial UI.

I'm going to tag this release, but we may scrap it depending on how we feel for now.

Basically, this ticket covers health metrics as reported by the runner. How the runner gathers the metrics is largely arbitrary and in the future may involve arbitrary data entirely. However, for now, the following metrics are reaped:

• cpu
• memory
• disk usage
• disk i/o

These items should be reported to the queuesvc for now and then be proxied to the datasvc for recording. The uisvc can retrieve but not set this data, and it will report the latest value recorded; allowing you to submit as frequently as you want metrics-wise.

For datasvc implementation, we will store a table of metrics with column->value pairs, where the columns are the above data. Again, a future implementation may do something more flexible and/or elaborate.

Each recording operation overwrites the previous recording operation. No log is kept in this version, this is to avoid having to write roll-up code for the metrics to avoid wasting space.

Keeping average values of the run is not in scope for this ticket, either. Instead, finished runs will report the last value given back by the runner. The runner can, of course, opt to record its own average metrics as the last metrics it sends, but this is not defined by this implementation.