processes are killed because of out of swap space

Mark Millard marklmi at yahoo.com
Sun Jan 5 22:07:31 UTC 2020



On 2020-Jan-5, at 12:41, Wojciech Puchar <wojtek at puchar.net> wrote:
> 
>>>> but still - there is A LOT of memory to be reclaimed. inactive is many gigabytes on my server.
>>>> 
>>>>> # Delay when persisstent low free RAM leads to
>>>>> # Out Of Memory killing of processes:
>>>>> vm.pageout_oom_seq=120
>>>> set to 300.
>> 
>> A good question is if changing this figure significantly
>> changes how long before the OOM kills significantly.
> 
> yes. but they are still killed.
> 
> the problem was geom_raid5. which i found not usable anyway - in case of system crash filesystem is in state of total mess.
> 
> I just decided to have less space with 2 gmirrors but at least proven, stable and fast.
> 
> All problems disappeared now.

Glad you isolated a way to avoid the problem in your
context.

If anyone is to work on geom_raid5, knowledge that:

A) it was getting OOM kills that the other technique did not
and:
B) vm.pageout_oom_seq changes were controlling when

likely would be useful background information. I
wonder if a bugzilla report of such would be
appropriate. (Some detail about the time frames
for the alternative vm.pageout_oom_seq figures might
be appropriate.)

===
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)



More information about the freebsd-hackers mailing list