Re: [pcp] PMDAs for lm_sensors, HDD SMART monitoringâ

[Ken McDonell <kenj@xxxxxxxxxxxxxxxx>]

On 02/01/16 22:07, David O'Shea wrote:
> ...
> Should I fork https://github.com/performancecopilot/pcp and file a pull
> request to improve the samples to refer to cpmda.PMDA_FETCH_*?
That will work.  Or send patches in email.

> I always err on the side of way too much detail in such diagrams, but in
> this case I'm personally interested just in what needs to happen in my
> PMDA.  Specifically, if a client is monitoring a few of my metrics, how
> does that appear to the PMDA, since I want to run 'smartctl' just once.
> However, that is a moot point now that I'm going to change the PMDA to
> have a separate thread doing the polling.
Too much detail is probably not what you wanted here ... the call graph 
through libpcp and libpcp_pmda can get pretty complicated for a request 
that arrives at a PMDA (which might be delivered as a PDU for a daemon 
PMDA, or a procedure call for a DSO PMDA).
Let me think about this for a while ...