From: Paul Turner <pjt@google.com>
To: Tejun Heo <tj@kernel.org>
Cc: Austin S Hemmelgarn <ahferroin7@gmail.com>,
Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@redhat.com>,
Johannes Weiner <hannes@cmpxchg.org>,
lizefan@huawei.com, cgroups <cgroups@vger.kernel.org>,
LKML <linux-kernel@vger.kernel.org>,
kernel-team <kernel-team@fb.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH 3/3] sched: Implement interface for cgroup unified hierarchy
Date: Mon, 24 Aug 2015 14:10:17 -0700 [thread overview]
Message-ID: <CAPM31R+i7BD8x9h_6wZVTa0zCB7XP0SGL5dSA-n6h9PTeAEhug@mail.gmail.com> (raw)
In-Reply-To: <20150824210223.GH28944@mtj.duckdns.org>
On Mon, Aug 24, 2015 at 2:02 PM, Tejun Heo <tj@kernel.org> wrote:
> Hello,
>
> On Mon, Aug 24, 2015 at 01:54:08PM -0700, Paul Turner wrote:
>> > That alone doesn't require hierarchical resource distribution tho.
>> > Setting nice levels reasonably is likely to alleviate most of the
>> > problem.
>>
>> Nice is not sufficient here. There could be arbitrarily many threads
>> within the hypervisor that are not actually hosting guest CPU threads.
>> The only way to have this competition occur at a reasonably fixed
>> ratio is a sub-hierarchy.
>
> I get that having hierarchy of threads would be nicer but am having a
> bit of difficulty seeing why adjusting priorities of threads wouldn't
> be sufficient. It's not like threads of the same process competing
> with each other is a new problem. People have been dealing with it
> for ages. Hierarchical management can be a nice plus but we want the
> problem and proposed solution to be justifiable.
Consider what happens with load asymmetry:
Suppose that we have 10 vcpu threads and 100 support threads.
Suppose that we want the support threads to receive up to 10% of the
time available to the VM as a whole on that machine.
If I have one particular support thread that is busy, I want it to
receive that entire 10% (maybe a guest is pounding on scsi for
example, or in the thread-pool case, I've passed a single expensive
computation). Conversely, suppose the guest is doing lots of
different things and several support threads are active, I want the
time to be shared between them.
There is no way to implement this with nice. Either a single thread
can consume 10%, and the group can dominate, or the group cannot
dominate and the single thread can be starved.
>
> Thanks.
>
> --
> tejun
next prev parent reply other threads:[~2015-08-24 21:10 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-08-03 22:41 [PATCHSET sched,cgroup] sched: Implement interface for cgroup unified hierarchy Tejun Heo
2015-08-03 22:41 ` [PATCH 1/3] cgroup: define controller file conventions Tejun Heo
2015-08-04 8:42 ` Peter Zijlstra
2015-08-04 14:51 ` Tejun Heo
2015-08-04 8:48 ` Peter Zijlstra
2015-08-04 14:53 ` Tejun Heo
2015-08-04 19:31 ` [PATCH v2 " Tejun Heo
2015-08-05 0:39 ` Kamezawa Hiroyuki
2015-08-05 7:47 ` Michal Hocko
2015-08-06 2:30 ` Kamezawa Hiroyuki
2015-08-07 18:17 ` Michal Hocko
2015-08-17 22:04 ` Johannes Weiner
2015-08-17 21:34 ` Johannes Weiner
2015-08-03 22:41 ` [PATCH 2/3] sched: Misc preps for cgroup unified hierarchy interface Tejun Heo
2015-08-03 22:41 ` [PATCH 3/3] sched: Implement interface for cgroup unified hierarchy Tejun Heo
2015-08-04 9:07 ` Peter Zijlstra
2015-08-04 15:10 ` Tejun Heo
2015-08-05 9:10 ` Peter Zijlstra
2015-08-05 14:31 ` Tejun Heo
2015-08-17 20:35 ` Tejun Heo
[not found] ` <CAPM31RJTf0v=2v90kN6-HM9xUGab_k++upO0Ym=irmfO9+BbFw@mail.gmail.com>
2015-08-18 4:03 ` Paul Turner
2015-08-18 20:31 ` Tejun Heo
2015-08-18 23:39 ` Kamezawa Hiroyuki
2015-08-19 16:23 ` Tejun Heo
2015-08-19 3:23 ` Mike Galbraith
2015-08-19 16:41 ` Tejun Heo
2015-08-20 4:00 ` Mike Galbraith
2015-08-20 7:52 ` Tejun Heo
2015-08-20 8:47 ` Mike Galbraith
2015-08-21 19:26 ` Paul Turner
2015-08-22 18:29 ` Tejun Heo
2015-08-24 15:47 ` Austin S Hemmelgarn
2015-08-24 17:04 ` Tejun Heo
2015-08-24 19:18 ` Mike Galbraith
2015-08-24 20:00 ` Austin S Hemmelgarn
2015-08-24 20:25 ` Tejun Heo
2015-08-24 21:00 ` Paul Turner
2015-08-24 21:12 ` Tejun Heo
2015-08-24 21:15 ` Paul Turner
2015-08-24 20:54 ` Paul Turner
2015-08-24 21:02 ` Tejun Heo
2015-08-24 21:10 ` Paul Turner [this message]
2015-08-24 21:17 ` Tejun Heo
2015-08-24 21:19 ` Paul Turner
2015-08-24 21:40 ` Tejun Heo
2015-08-24 22:03 ` Paul Turner
2015-08-24 22:49 ` Tejun Heo
2015-08-24 23:15 ` Paul Turner
2015-08-25 2:36 ` Kamezawa Hiroyuki
2015-08-25 21:13 ` Tejun Heo
2015-08-25 9:24 ` Ingo Molnar
2015-08-25 10:00 ` Peter Zijlstra
2015-08-25 19:18 ` Tejun Heo
2015-08-24 20:52 ` Paul Turner
2015-08-24 21:36 ` Tejun Heo
2015-08-24 21:58 ` Paul Turner
2015-08-24 22:19 ` Tejun Heo
2015-08-24 23:06 ` Paul Turner
2015-08-25 21:02 ` Tejun Heo
2015-09-02 17:03 ` Tejun Heo
2015-09-09 12:49 ` Paul Turner
2015-09-12 14:40 ` Tejun Heo
2015-09-17 14:35 ` Peter Zijlstra
2015-09-17 14:53 ` Tejun Heo
2015-09-17 15:42 ` Peter Zijlstra
2015-09-17 15:10 ` Peter Zijlstra
2015-09-17 15:52 ` Tejun Heo
2015-09-17 15:53 ` Peter Zijlstra
2015-09-17 23:29 ` Tejun Heo
2015-09-18 11:27 ` Paul Turner
2015-10-01 18:46 ` Tejun Heo
2015-10-15 11:42 ` Paul Turner
2015-10-23 22:21 ` Tejun Heo
2015-10-24 4:36 ` Mike Galbraith
2015-10-25 2:18 ` Tejun Heo
2015-10-25 3:43 ` Mike Galbraith
2015-10-27 3:16 ` Tejun Heo
2015-10-27 5:42 ` Mike Galbraith
2015-10-27 5:46 ` Tejun Heo
2015-10-27 5:56 ` Mike Galbraith
2015-10-27 6:00 ` Tejun Heo
2015-10-27 6:08 ` Mike Galbraith
2015-10-25 3:54 ` Linus Torvalds
2015-10-25 9:33 ` Ingo Molnar
2015-10-25 10:41 ` Theodore Ts'o
2015-10-25 10:47 ` Florian Weimer
2015-10-25 11:58 ` Theodore Ts'o
2015-10-25 13:17 ` Florian Weimer
2015-10-25 13:40 ` Getrandom wrapper Theodore Ts'o
2015-10-26 13:32 ` Florian Weimer
2015-10-26 14:10 ` [PATCH 3/3] sched: Implement interface for cgroup unified hierarchy Peter Zijlstra
2015-08-04 19:32 ` [PATCH v2 " Tejun Heo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAPM31R+i7BD8x9h_6wZVTa0zCB7XP0SGL5dSA-n6h9PTeAEhug@mail.gmail.com \
--to=pjt@google.com \
--cc=ahferroin7@gmail.com \
--cc=akpm@linux-foundation.org \
--cc=cgroups@vger.kernel.org \
--cc=hannes@cmpxchg.org \
--cc=kernel-team@fb.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lizefan@huawei.com \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=tj@kernel.org \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).