ubuntu-phone team mailing list archive

[Ted Gould <ted@xxxxxxxxxx>]

On Mon, 2013-06-10 at 10:17 +0200, Michael Zanetti wrote:

> I've been thinking about some small daemon that would collect CPU times of 
> apps and send that to some stats server in an automated manner. I think that 
> would be very useful. Shouldn't be too hard to create some snapshots every x 
> seconds and send collected reports to a server once a day or so. There we 
> could create stats on which binaries use the most CPU and get a good feeling 
> where optimisation is needed. In a second step this could also include crash 
> traces and such.


We have the infrastructure already set up for this with Whoopsie and
errors.ubuntu.com.  We can give it reports and it will send them
asynchronously, for instance, when there is networking.  It can handle
anything from reporting data, to recoverable errors, to crashes.  I do
think we should use it more, and we started a small project to make it
easier to do metrics collections, though it's not complete yet:

https://launchpad.net/libmetrics

For information on recoverable errors you can look at:

/usr/share/apport/recoverable_problem 

We're working on a patch for GLib so that everytime a g_critical() is
printed a recoverable error is reported.  That way we can start tracking
and prioritizing these errors.

Ted

Attachment: signature.asc
Description: This is a digitally signed message part