Resource consumption for indexing user data, especially files and mails

[Martin Steigerwald]

Cc'ing you as you cc'd me. I have no need to be cc'd.

Cornelius Schumacher - 26.06.21, 13:26:11 CEST:
> The measurement procedures I described in my talk only measure total
> resource usage of the overall system. They don't try to tie it to
> specific processes or differentiate between the application itself and
> any background services. They do measure different usage profile with
> the application running through a realistic usage scenario, a
> scenario where the application is idling, and a baseline with the
> application not installed or running.

So they eventually pick up background indexing activity. However, I 
wonder whether the amount of user data has been comparable to that of a 
typical user. I consider myself to be a heavy user with tons of files and 
mails. But I have no idea what the average usage might be.

> It would be interesting to do measurements comparing different
> settings to the degree that's technically feasible. So comparing
> energy consumption with indexing enabled and disabled would show the
> difference and also give a number how big the part of indexing is in
> the overall energy consumption.
> 
> A tricky part might be to get background processes controlled in a way
> to get reproducible measurements.

That. I never quite got when exactly they will start indexing.

For baloo either "balooctl purge" or "balooctl check" or related DBUS 
methods could work.

For Akonadi you could remove ~.local/akonadi/search_db. However I do not 
know how to trigger the indexing run then. It eventually will do it, but 
I never got how to predict when.

At the current time I do not see myself getting up to speed with C++ – 
my C coding skills already are pretty dated and I never really learned 
C++ – and the code base enough to fix any of that myself, but I am 
willing to help with measurements and testing as time permits.

Best,
-- 
Martin