On Fri, 10 Jul 2020 at 21:28, Yafang Shao <laoar.shao@xxxxxxxxx> wrote:
>
> Recently we found an issue on our production environment that when memcg
> oom is triggered the oom killer doesn't chose the process with largest
> resident memory but chose the first scanned process. Note that all
> processes in this memcg have the same oom_score_adj, so the oom killer
> should chose the process with largest resident memory.
>
> Bellow is part of the oom info, which is enough to analyze this issue.
> [7516987.983223] memory: usage 16777216kB, limit 16777216kB, failcnt 52843037
> [7516987.983224] memory+swap: usage 16777216kB, limit 9007199254740988kB, failcnt 0
> [7516987.983225] kmem: usage 301464kB, limit 9007199254740988kB, failcnt 0
[...]
> [7516987.984221] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=3aa16c9482ae3a6f6b78bda68a55d32c87c99b985e0f11331cddf05af6c4d753,mems_allowed=0-1,oom_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184,task_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184/1f246a3eeea8f70bf91141eeaf1805346a666e225f823906485ea0b6c37dfc3d,task=pause,pid=5740,uid=0
> [7516987.984254] Memory cgroup out of memory: Killed process 5740 (pause) total-vm:1028kB, anon-rss:4kB, file-rss:0kB, shmem-rss:0kB
> [7516988.092344] oom_reaper: reaped process 5740 (pause), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
>
> We can find that the first scanned process 5740 (pause) was killed, but its
> rss is only one page. That is because, when we calculate the oom badness in
> oom_badness(), we always ignore the negtive point and convert all of these
> negtive points to 1. Now as oom_score_adj of all the processes in this
> targeted memcg have the same value -998, the points of these processes are
> all negtive value. As a result, the first scanned process will be killed.
>
> The oom_socre_adj (-998) in this memcg is set by kubelet, because it is a
> a Guaranteed pod, which has higher priority to prevent from being killed by
> system oom.
>
> To fix this issue, we should make the calculation of oom point more
> accurate. We can achieve it by convert the chosen_point from 'unsigned
> long' to 'long'.
>
> [cai@xxxxxx: reported a issue in the previous version]
> [mhocko@xxxxxxxx: fixed the issue reported by Cai]
> [mhocko@xxxxxxxx: add the comment in proc_oom_score()]
> Signed-off-by: Yafang Shao <laoar.shao@xxxxxxxxx>
> Acked-by: Michal Hocko <mhocko@xxxxxxxx>
> Cc: David Rientjes <rientjes@xxxxxxxxxx>
> Cc: Qian Cai <cai@xxxxxx>
>
> ---
> v2 -> v3:
> - fix the type of variable 'point' in oom_evaluate_task()
> - initialize oom_control->chosen_points in select_bad_process() per Michal
> - update the comment in proc_oom_score() per Michal
>
> Signed-off-by: Yafang Shao <laoar.shao@xxxxxxxxx>
Tested-by: Naresh Kamboju <naresh.kamboju@xxxxxxxxxx>
I have noticed kernel panic with v2 patch while running LTP mm test suite.
[ 63.451494] Out of memory and no killable processes...
[ 63.456633] Kernel panic - not syncing: System is deadlocked on memory
Then I have removed the v2 patch and applied this below v3 patch and re-tested.
No regression noticed with v3 patch while running LTP mm on x86_64 and arm.
OTOH,
oom01 test case started with 100 iterations but runltp got killed after the
6th iteration [3]. I think this is expected.
test steps:
- cd /opt/ltp
- ./runltp -s oom01 -I 100 || true
[ 209.052842] Out of memory: Killed process 519 (runltp)
total-vm:10244kB, anon-rss:904kB, file-rss:4kB, shmem-rss:0kB, UID:0
pgtables:60kB oom_score_adj:0
[ 209.066782] oom_reaper: reaped process 519 (runltp), now
anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
/lava-1558245/0/tests/0_prep-tmp-disk/run.sh: line 21: 519 Killed
./runltp -s oom01 -I 100
> ---
> fs/proc/base.c | 11 ++++++++++-
> include/linux/oom.h | 4 ++--
> mm/oom_kill.c | 22 ++++++++++------------
> 3 files changed, 22 insertions(+), 15 deletions(-)
Reference test jobs,
[1] https://lkft.validation.linaro.org/scheduler/job/1558246#L9189
[2] https://lkft.validation.linaro.org/scheduler/job/1558247#L17213
[3] https://lkft.validation.linaro.org/scheduler/job/1558245#L1407
