[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#851471: linux-image-4.8.0-2-686-pae: premature oom killer invocation running 4.8.15



Control: tag -1 moreinfo

On Sun, 2017-01-15 at 11:05 +0100, Florian Siegesmund wrote:
> Package: src:linux
> Version: 4.8.15-2
> Severity: critical
> Tags: upstream
> Justification: breaks the whole system
> 
> Dear Maintainer,
> 
> since running 4.8.x kernel from testing, the system crashes each morning
> soon after anacron starts cron.daily job. This is obviously caused by oom
> killer kicking in while hardly any swap used at all. This situation
> persists for several revisions of the package.
> 
> When falling back to linux-image-3.16.0-4-686-pae (3.16.36-1+deb8u2
> currently) everything is fine again.
> 
> This seems to be a known issue with kernel 4.8.x, e.g.:
> 
> https://marius.bloggt-in-braunschweig.de/2016/11/17/linuxkernel-4-74-8-und-der-oom-killer/
> https://unix.stackexchange.com/questions/300106/why-is-the-oom-killer-killing-processes-when-swap-is-hardly-used
> 
> It seems to me that commit 7838fbe25a95ce2cd6e8ae27a76d369365da89d4
> (kernel ChangeLog-4.8.12) was meant to deal with this problem, but it
> doesn't work for me.
>
> I will add a sample syslog for this issue.

That includes an RCU error before the OOM.  Was that the first message
relating to RCU?

Linux 4.9 is said to have better OOM behaviour and is now available in
unstable.  Can you test whether it works better on this system?

Ben

> I have yet to test the "sync && echo 1 > /proc/sys/vm/drop_caches"
> workaround mentioned in the web references or 4.9.x from unstable.

-- 
Ben Hutchings
Hoare's Law of Large Problems:
        Inside every large problem is a small problem struggling to get
out.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: