-
Notifications
You must be signed in to change notification settings - Fork 48
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
nohang-desktop stopped working? #142
Comments
I can't say that nohang has stopped working. I can assume that kernel's OOM killer triggered prematurely. We need to find out why. |
Dec 23 12:13:22 gentoo nohang-desktop[81157]: PSI: 7.95 | MemAvail: 9149 M, 39.0 % | SwapFree: 654 M, 11.1 % then Dec 23 12:13:33 gentoo kernel: kswapd0 invoked oom-killer: |
Maybe try min_mem_report_interval = 0 to check (and log) memory more often. |
Is MGLRU enabled? With MGLRU enabled, the killer may be more aggressive. |
I reproduced this with MGLRU: with nohang disabled, killer killed with a lot of SwapfFree even with min_ttl_ms=0. |
MGLRU was enabled. Disabled MGLRU with |
Warnings are triggered but no action is taken.
|
You can log each memory check: min_mem_report_interval = 0 You can set more aggressive settings, increase the frequency of memory checks: fill_rate_mem = 8000 |
Question: is it nohang missing the state when it needs to react, or is it the kernel killing the process prematurely, before the state at which nohang starts to react. |
I see that nohang understands that there is a problem ongoing, but it doesn't react fast enough. OOM kills it first.
|
fill_rate_mem = 100 It's too low! It can react in time if the memory fills up at a speed of up to 100 MiB/s, but not more (but checks beetween min_sleep and max_sleep). |
Increased the fill_rate, but no change :( |
This shouldn't have happened. This command does not cause an OOM, it terminates itself if no one interferes with it.
|
|
Tried without any swap, same results.
I don't use zfs, ext4 and btrfs.
I use 6.12.5 now.
It seems that oomkiller is triggered earlier than usual. Maybe cachy
patches have something to do with it.
Στις Τρί, 24 Δεκ 2024, 13:05 ο χρήστης Alexey Avramov <
***@***.***> έγραψε:
…
- Try after swapoff
- What is your fs? (ZFS uses Adaptive Replacement Cache)
- What happens when other kernel versions used
—
Reply to this email directly, view it on GitHub
<#142 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABNOHTITMG42DCT7R7GVR7L2HE5YPAVCNFSM6AAAAABUCZ5AM6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNRQHE4TMMBUGI>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
Hi, I use cachy-os kernel patches & zram and I have noticed that nohang-desktop is not protecting from OOM anymore.
kswapd0 is triggering oom-killer.
How can I debug this further? Can I change some limits in nohang-desktop to force nohang to intervene earlier?
Logs:
Config:
The text was updated successfully, but these errors were encountered: