[systemd-devel] [RFC] core: introduce ExitOnIdle= and ExitOnIdleSec=

[Martin Pitt]

Hello Kyungmin,

Kyungmin Park [2015-04-21  9:21 +0900]:
> At mobile. some daemon is not doing exact daemon task. it acts like
> app. so it's kill-able based on priority. now it can't know it's idle
> or not. In the app-like daemon developer, they don't want to exit
> since performance reason. but in the view of system admin, it's
> resource waste. it's better to kill if it know it's idle.

I'm not at all convinced that this is a good strategy. For maximum
efficiency you *do* want to let apps stay in RAM/in the background as
long as you have enough memory, so that they don't need to
re-load/link/initialize when you call them again. So regular
time-based sweeps are too aggressive here and lead to unnecessary app
starts (which take a lot more CPU/battery than a simple context
switch).

On the other hand, if you start a new app and it doesn't get enough
RAM, you want/need to start killing existing background apps
oldest-to-newest (or perhaps biggest-to-smallest) until you have
enough RAM again, but that needs to happen immediately. You don't want
to wait until the next idle counter timer for this to happen.

Finally, waking up the device every 30 s to clean up "idle" apps takes
CPU/battery by itself.

So I see no use case for idle timer based cleanup. Can you please
explain why they are better than on-demand cleanup?

Thanks,

Martin
-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.freedesktop.org/archives/systemd-devel/attachments/20150421/3cd29c5e/attachment.sig>