oomk kills too much apps

asked 2016-12-05 01:13:55 +0200

cemoi71 gravatar image

updated 2016-12-05 01:18:45 +0200

Hello,

i remark shortly that the oom-killer function of the system stop too much apps.

For example if i take a native browser like the official one or webcat.
And try to load all the iso file from this website: http://iso.linuxquestions.org/

then later, the wlan connection will be unstable, all of of interruption will be done. Then all the active app will crash. The covers still remain in home, but by taping on it, the app will be recall.

Firstly, from this oomk function, i would expect that it kill first the app which take the most place in ram.
Or maybe that it kills after a specific "survival" priority schema.
I can represent my self that a webbrowser would have a low priority. And if it takes too much ram, it could be killed first alone, before other one.
Then, maybe it's a kind of speculation from me, but i can't imagine that something with this function disturbed the wlan connection.
It could be an other bug, but each time i try to load a lots of file, the wlan connection goes interrupted, until all the apps will crash.
It's for me too difficult not to associate it, with the oomk, there is for me a kind of bug behind.

I find actually that the oomk action is a little bit too rough.
Would it be possible to optimize/improve it please?
Maybe analyse what's happened on the wlan at the moment of an high dataloading?

I have a jp1, with last official release.

note: i remark now that after my experiment with the official browser, that this one could never start again.
After start it, the cover is working on it, and after a while it disappear again without to have the browser-app.

Have a nice sail.

edit retag flag offensive close delete

Comments

Aggre. It needs improvements, and why not the possibility to set it off, but it's not Jolla like ..

malibu1106 ( 2016-12-05 03:29:53 +0200 )edit
3

Really, you didnt found any existing questions about oom and created new one?

coderus ( 2016-12-05 04:45:40 +0200 )edit
2

If you do a search on "oom killer" you will find several articles about this, for example this one which has some very clear information.

filipb92 ( 2016-12-05 10:05:02 +0200 )edit

You didn't understand the issue correctly. The problem here is that downloading big files with the UI is eating too much RAM. It's a know issue for more than 2 years: https://together.jolla.com/question/48921/lipstick-eating-lot-of-ram-while-downloading-big-files-via-stock-browser/

Then of course, all other processes get killed. I think you can close it as duplicate.

Sthocs ( 2016-12-05 11:59:23 +0200 )edit

@coderus, please if you think that i didn't search for a question which explicit ask for improvement with exact experiences as fact about it, then i've nothing to do in our mature world.
Your question seems to me that on your side you didn't make the search job too, otherwise you would gave me one thread that maybe i didn't see.
But I spent for this more than one week for it, between testing different conditions and look after reproducibility, and try to find an adapted thread which explicit ask for improvement (sorry i'm just human).
I have this issue since more than 6 months sometimes non concretely by native apps, and more often with android apps. Until i discovered shortly this oomk function [here].(https://together.jolla.com/question/152443/what-happened-and-crash-behind-the-gui/) Better constructive is not ask questions which can have only a dead end response.
I think you too have actually something better to do. right? otherwise you may want help.
I think my thread present concrete facts, and ask for something concrete.
If you know an other concrete thread, please you may help by pointing on it, i have no problem.
But i find such a question as you did, is just condescend and not respectful.
So should we struggle unnecessary and close the eye on the problem, or make it for the best together?
Many thanks for understanding

cemoi71 ( 2016-12-05 12:19:44 +0200 )edit