-
Notifications
You must be signed in to change notification settings - Fork 149
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Option to ignore oom_score completely #140
Comments
Hi! What kind of applications are we talking about here? |
Well, in my case these are firefox, intellij idea, lutris (wine launcher), etc. Actually, whenever I open htop with sort by RES, I'm seeing a big wall of zeroes in OOM column, except for several rather small applications. I can figure out that this behavior isn't normal, but I really can't see for a solution other than, for example, forcefully set adj of all processes to zero by cron, which is surely somewhat clumsy approach. |
Huh, strange! Can you maybe post a screenshot of what you get in htop?
…On Mon, Aug 19, 2019 at 2:47 AM AbsurdlySuspicious ***@***.***> wrote:
Well, in my case these are firefox, intellij idea, lutris (wine launcher),
etc. Actually, whenever I open htop with sort by RES, I'm seeing a big wall
of zeroes in OOM column, except for several rather small applications. I
can figure out that this behavior isn't normal, but I'm really can't see
for a solution other than, for example, forcefully set adj of all processes
to zero by cron, which is surely somewhat clumsy approach.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#140?email_source=notifications&email_token=AACGA753HOGANFEUW6BLSEDQFHURPA5CNFSM4IJ3NJ22YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD4RM5CI#issuecomment-522374793>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AACGA77RUBVB3CHVHUWSG6LQFHURPANCNFSM4IJ3NJ2Q>
.
|
These applications do not set @AbsurdlySuspicious Please run https://github.com/hakavlad/nohang/blob/master/oom-sort and print output here. This script prints also |
sorted ooms: https://gist.github.com/AbsurdlySuspicious/2cc0988c5261a1ff595b4989c9a8d04e (truncated) |
@AbsurdlySuspicious Seems like you set OOMScoreAdjust=-1000 in user.slice or in session.scope. I'd like to see your |
@AbsurdlySuspicious Try also:
It will display units with modified oom_score_adj.
|
@hakavlad grep gives me *.service matches only (dbus, dm-event, etc). I'll try running nohang slightly later |
Please show all output. |
Nothing related to slices or scopes, just like i said |
My guess: Ananicy does this, see https://github.com/Nefelim4ag/Ananicy/blob/6181b1dd3954c659eae905f0320c9c31562ecdb6/ananicy.py#L228 . |
Also, looking at https://gist.github.com/AbsurdlySuspicious/2cc0988c5261a1ff595b4989c9a8d04e , ananicy is the first (lowest PID) process that has a -999 value for oom_score_adj. (other than |
You're right, it was ananicy, I've disabled oom_score_adj applying and everything is fine now. Thank you guys for helping me with this. I'd still like you to consider keeping this as low-prio feature request, as it still can be helpful in some cases, even if they're caused by mistake. |
In my case there's a lot of memory-eager applications which sets themselves
oom_score_adj=-1000
and therefore I keep getting xfce4-notifyd (with 25M VmRSS) killed first which occasionally gets an oom_score value of 1 instead of the process that actually ate all of my ram and triggered -m limits (but keeps its oom_score at 0).I don't really know is that a common problem but anyway, in my opinion it'll be good to have an option to decide which process to kill based exclusively on VmRSS value
The text was updated successfully, but these errors were encountered: