Re: [pcp] PCP versions of console tools - Next Steps

[Nathan Scott <nathans@xxxxxxxxxx>]

Hi guys,

----- Original Message -----
> On 12/05/16 03:25, Sitaram Shelke wrote:
> > Thank you for the input and yes using pmFetchGroup() is the next task on
> > my board. Ryan suggested trying pmFetch() first and then going after
> > pmFetchGroup().
> 
> Thanks Sitaram, I've checked it out and it looks like you've got the
> hang of the fetchgroup API.
> 
> I've added a task to check out the pmcc module and some of the classes
> in there that abstracts the pmapi further. It's also got some useful
> classes for building tools like pidstat.

Best bet would be pmcc module extension to allow use of fetchgroups under
the hood - noone has taken on this work yet though.

Ultimately what we'd like is for python scripts to be just: command line
options (i.e. a pmOptions object) plus report() function, and the helper
"convenience" classes take care of the rest (building on top of whichever
APIs they need to under the covers).

> I've added more information in the Wekan card but have a look at
> pcp-iostat.py as a starting point. It's likely we will want to use this
> for the implementation of pidstat.

+1  - and feel free to extend the python modules as part of your work too
Sitaram - there's definitely room for improvement there still.

cheers.

--
Nathan