Re: [RFC PATCH 0/3] mm: memcontrol: delayed force empty

[Greg Thelen]

Yang Shi <yang.shi@xxxxxxxxxxxxxxxxx> wrote:

> On 1/3/19 11:23 AM, Michal Hocko wrote:
>> On Thu 03-01-19 11:10:00, Yang Shi wrote:
>>>
>>> On 1/3/19 10:53 AM, Michal Hocko wrote:
>>>> On Thu 03-01-19 10:40:54, Yang Shi wrote:
>>>>> On 1/3/19 10:13 AM, Michal Hocko wrote:
>> [...]
>>>>>> Is there any reason for your scripts to be strictly sequential here? In
>>>>>> other words why cannot you offload those expensive operations to a
>>>>>> detached context in _userspace_?
>>>>> I would say it has not to be strictly sequential. The above script is just
>>>>> an example to illustrate the pattern. But, sometimes it may hit such pattern
>>>>> due to the complicated cluster scheduling and container scheduling in the
>>>>> production environment, for example the creation process might be scheduled
>>>>> to the same CPU which is doing force_empty. I have to say I don't know too
>>>>> much about the internals of the container scheduling.
>>>> In that case I do not see a strong reason to implement the offloding
>>>> into the kernel. It is an additional code and semantic to maintain.
>>> Yes, it does introduce some additional code and semantic, but IMHO, it is
>>> quite simple and very straight forward, isn't it? Just utilize the existing
>>> css offline worker. And, that a couple of lines of code do improve some
>>> throughput issues for some real usecases.
>> I do not really care it is few LOC. It is more important that it is
>> conflating force_empty into offlining logic. There was a good reason to
>> remove reparenting/emptying the memcg during the offline. Considering
>> that you can offload force_empty from userspace trivially then I do not
>> see any reason to implement it in the kernel.
>
> Er, I may not articulate in the earlier email, force_empty can not be 
> offloaded from userspace *trivially*. IOWs the container scheduler may 
> unexpectedly overcommit something due to the stall of synchronous force 
> empty, which can't be figured out by userspace before it actually 
> happens. The scheduler doesn't know how long force_empty would take. If 
> the force_empty could be offloaded by kernel, it would make scheduler's 
> life much easier. This is not something userspace could do.

If kernel workqueues are doing more work (i.e. force_empty processing),
then it seem like the time to offline could grow.  I'm not sure if
that's important.

I assume that if we make force_empty an async side effect of rmdir then
user space scheduler would not be unable to immediately assume the
rmdir'd container memory is available without subjecting a new container
to direct reclaim.  So it seems like user space would use a mechanism to
wait for reclaim: either the existing sync force_empty or polling
meminfo/etc waiting for free memory to appear.

>>>> I think it is more important to discuss whether we want to introduce
>>>> force_empty in cgroup v2.
>>> We would prefer have it in v2 as well.
>> Then bring this up in a separate email thread please.
>
> Sure. Will prepare the patches later.
>
> Thanks,
> Yang