Re: Runaway process and oom-killer

[John Sigler <linux.kernel@free.fr>]

Chris Friesen wrote:

> Helge Hafting wrote:
> 
>> My guess:
>> Something needs memory but finds there is none to be had
>> oom-killer is invoked and targets myapp.
>> myapp takes some time to die. Particularly, the memory it uses
>> isn't freed up instantly.
> 
> Has anyone considered actually bumping up the priority of the task being 
> killed so that it gets to run and free up its resources in a timely manner?

On this system,
myapp runs in SCHED_RR with priority 80.
IRQ handlers run in SCHED_FIFO with priority 50.

# ps -eo comm,class,rtprio,ni,pri --sort -rtprio
COMMAND         CLS RTPRIO  NI PRI
posix_cpu_timer FF      99   - 139
myapp           RR      80   - 120
softirq-high/0  FF      50   -  90
softirq-timer/0 FF      50   -  90
softirq-net-tx/ FF      50   -  90
softirq-net-rx/ FF      50   -  90
softirq-block/0 FF      50   -  90
softirq-tasklet FF      50   -  90
softirq-sched/0 FF      50   -  90
softirq-hrtimer FF      50   -  90
softirq-rcu/0   FF      50   -  90
IRQ-7           FF      50   -  90
IRQ-8           FF      50   -  90
IRQ-14          FF      50   -  90
IRQ-12          FF      50   -  90
IRQ-1           FF      50   -  90
IRQ-10          FF      50   -  90
IRQ-11          FF      50   -  90
IRQ-5           FF      50   -  90
IRQ-3           FF      50   -  90
IRQ-4           FF      50   -  90
events/0        FF       1   -  41
init            TS       -   0  24
desched/0       TS       - -10  34
khelper         TS       -  -5  29
kthread         TS       -  -5  27
kblockd/0       TS       -  -5  21
kacpid          TS       -  -5  19
kseriod         TS       -  -5  29
pdflush         TS       -   0  17
pdflush         TS       -   0  24
kswapd0         TS       -  -5  23
flush_filesd/0  TS       -  -5  29
aio/0           TS       -  -5  22
syslogd         TS       -   0  21
klogd           TS       -   0  21
sshd            TS       -   0  21
acpid           TS       -   0  16
agetty          TS       -   0  24
agetty          TS       -   0  21
agetty          TS       -   0  21
agetty          TS       -   0  21
[...]

How do the scheduling class and priority of the process come into play 
when the kernel comes to reclaim memory after the oom-killer has decided 
to snipe that particular process?

> We've done some experimenting with actually putting it in SCHED_RR and 
> it seems to help (in the case of other busy SCHED_RR tasks on the 
> system).  Admittedly we have an older kernel, so behaviour may be 
> different now.

Thanks for sharing your experience.

Regards.