I tested the halving briefly, and had similar results, unfortunately.
It looks like the limits are quite well tuned for X10II. If that is indeed the case, I see no reason to double the values for X10III only because it has double the RAM.
We are slowly narrowing the acceptable window here, which is good 
Itās been two weeks since Iāve been using halved values on the 10 III, without any problems and with not a single occurrence of closing any app due to OOM, with the single exception of the web browser, which happened once. So Iām more than satisfied.
My experience showed better results when halving the freemem values on my X10II. With the default settings, the Android App Support system gets killed most of the time after I watch a video in Fennec (Androids Firefox Browser), as soon as I stop using my device. Next time I turn on my screen, all Android apps are greyed out and opening one of them starts the Android App support first.
Since I started adjusting the freemem values, this rarely happened.
As Iām only using a few apps on my phone, this may not be the average user use case.
joining the boat here just tweaked minfree on my 10 II with 58580,70100,81620,93140,128245, I have 4 Android apps open + stock browser + Fernschreiber, and the system is still smooth, and no crash so far 
So I just changed my default values after I really got sick with apps closing down randomly. Following @cartronās tweak I changed from
[root@Xperia10II-DualSIM defaultuser]# cat /sys/modul
e/lowmemorykiller/parameters/minfree
117160,140200,163240,186280,256490
to
[defaultuser@Xperia10II-DualSIM ~]$ cat /sys/mod
ule/lowmemorykiller/parameters/minfree
58580,70100,81620,93140,128245
I will report back in the next couple of days whether I experience an improvement or not.
What I learned so far:
I have also modified the config on my 10 II.
Here is my experience:
I am back to the default configuration.
I was hit by random OOM kills, too that made using the device (10 Mark 2) nearly impossible.
There where a few things I recognized:
SailfishOS browser pinch-to-zoom eats memory, e.g. open some github project source code file and zoom in. Now scroll a little bit inside that zoomed file and bang there you go (LMKD is killing a bunch of running applications in a blink)
I had a memory leak caused by faulty / broken Sailfish MTA that consumed a lot of memory but alas itās a system process so other running non-system applications got killed first
My solution to the latter issue was to re-create my mail account which resolved the MTA system process memory leak that caused LMKD to hunt down my other non-system applications.
Now thatās what I call āinterestingāā¦ How did you figure it out that it was the MTA process that had the memory leak? Asking for a friend 
To be honest I had a bad feeling when syncing my mailbox (IMAP4) suddenly took very long and ended eventually with sync failures. After that I recognized that running multiple applications (Android: ntfy.sh, Schildichat; SFOS: Mail (backround), browser, Quickdit, forum viewer, Podqast) where killed overnight.
I didnāt dig any deeper nor analysed logs (which would be more objective). However after I fixed my broken mail account no overnight or common use crashes of running the same set of applications occured to me (knocking on wood).
What I wanted to say is that system processes are allowed higher memory demand before they are terminated so in the end faulty system processes can trigger the LMKD killing of unprivileged processes and you wonder why the heck are my applications crashing all the time.
I updated to 4.4.0.72 and my changes to the /vendor/etc files was reverted. Just realized when I got pulseaudio killed today (or yesterday).
yes, this doesnāt survive a reboot.
Iām looking at options, it looks like using udevadm could be an option, iāll try to find some time over the week-end to look at that, and will report back.
typo in your formula
it is *4 / 1024
You can edit this file, it will work till the next OS update. The numbers that direc posted last were the best I think, or values used by 10 II, both quite similar.
omg was to stupid to click on insert in vi
LOL
thanks
Thereās place for improvement on the XA2 devices tooā¦ Iāve had more frequent crashes lately, seemingly related to RAM usage. Typically, my browser would be killed when starting a big Android app. Or the Android app would fail to start if there were too many apps open.
Hereās what helped on my XA2:
printf 768,1024,2048,4096 >/
sys/module/lowmemorykiller/parameters/minfree
Iāve been using these settings for some weeks. Two other XA2 users have also tried them successfully. There still are OOM kills, but markedly fewer.
To recap, on Sony 10 Xperia III, we are changing the default value of āminfreeā (which killer was that?) by:
[root@Xperia10III defaultuser]# nano /vendor/etc/init/init.lena.rc
167160,190200,213240,236280,356490
to be changed to:
65536,73728,90112,131072,196608
Save and Reboot.
A Reboot will reset the values to default. How to overcome that?
Do we change the values from here instead?
[root@Xperia10III defaultuser]# nano /vendor/etc/init/init.lena.rc
# LMK tunning
write /sys/module/lowmemorykiller/parameters/enable_lmk 1
write /sys/module/lowmemorykiller/parameters/minfree "167160,190200,213240,236280,356490"
write /sys/module/lowmemorykiller/parameters/adj "0,58,147,529,1000"
write /sys/module/lowmemorykiller/parameters/vmpressure_file_min 105984
write /sys/module/lowmemorykiller/parameters/oom_reaper 1
Yes indeed, survives a reboot.
Is this the only step or change that we need to perform, changing the values within ā/vendor/etc/init/init.lena.rcā; or there is another step need to be taken?
Like
[root@Xperia10III defaultuser]# nano /proc/sys/vm/swappiness
And chang the value from 25 to 50?
50
Yet that resets back to value 25, after a reboot.
And regarding
Are these values, ā15360,19200,23040,26880,34415,43737ā imply to Sony 10 Xperia III, or Sony 10 Xperia II ?
My Sony 10 Xperia III is in constant freeze and crash when I do have any Android app up, specifically OSMAnd android map app, or any mapping package, like Pure Maps.
This and for some reason only happens when I am outdoors, walking within the city.
Perhaps this RAM process of killing apps what is causing this, and with the GPS issues, networking [since I reside in the US, not all bands my phone can pickup]?
Thank you all for this wealth of informationā¦ One of the best forums to follow
Those are Xperia 10 II values that I suggested using on 10 III. Those are not that much lower than what 10 III is using, so should be pretty safe.
What direc suggested should bring bigger improvements though, 65536,73728,90112,131072,196608 as you even said in your post.
Thank you for the update, oh, I stole all the info and and compiled a summary; I am thankful to all of you for the this great troubleshooting.
Regarding āpage fileā, after a reboot the value on the phone will set back to default, on Sony 10 Xperia III, when I use this command, after SSH into the phone:
nano /proc/sys/vm/swappiness
I do get the value (which I assume in percentage 25% of total RAM of 6GB)
25
Of course, I do change the āpercentageā value to
50
And a reboot will set the value back to ā25ā:
cat /proc/sys/vm/swappiness
25
I have tried different methods, thinking the page file is somehow controlled by ZRAM settings, like at:
cat /vendor/etc/fstab.pdx213
and at this line:
#ZRAM/SWAP
/dev/block/zram0 none swap defaults zramsize=1073741824,max_comp_streams=8
the value of 1073741824 = 1024MB
And when I replaced these numbers ā1073741824ā with 3GB, and did reboot, I do not find the value changed within ā/proc/sys/vm/swappinessā, rather, disapperance of ZRAM swap value, and even reverting back, I do get the same result below [I sespuct the way how I did edit the file]
cat sys/block/zram0/disksize
will yield at the moment,
[root@Xperia10III defaultuser]# cat sys/block/zram0/disksize
cat: can't open 'sys/block/zram0/disksize': No such file or directory
Yet, would it be from there, sys/block/zram0/disksize, the value of ZRAM swap page-file size will be defined, and persistent even after a reboot?
Do I need to recreate the ZRAM via dd command?"
Thank you
Setting the values in /vendor/etc/init/init.lena.rc indeed survive reboot; thatās what I currently use as well.
Swappiness value is set at boot time, it seems to be set in /vendor/hw/init.pdx213.rc but I havenāt tested that. Try it on your own risk! But; I donāt think swappiness has that much effect on all this, I suggest leaving that alone. OTOH it doesnāt hurt either, so thereās that.
I havenāt tried increasing zram size, but I think 3GB is way too much. If the kernel canāt allocate that much, you wonāt have zram at all. And if it succeeded, and almost all of it was in use, youād have serious trashing going on and the system would have grinded to a halt long before that.
You donāt have to create zram via dd, the kernel handles it all.