All of lore.kernel.org
 help / color / mirror / Atom feed
From: Mel Gorman <mgorman@techsingularity.net>
To: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Phil Auld <pauld@redhat.com>, Hillf Danton <hdanton@sina.com>,
	Ingo Molnar <mingo@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Juri Lelli <juri.lelli@redhat.com>,
	Dietmar Eggemann <dietmar.eggemann@arm.com>,
	Steven Rostedt <rostedt@goodmis.org>,
	Ben Segall <bsegall@google.com>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Parth Shah <parth@linux.ibm.com>,
	Valentin Schneider <valentin.schneider@arm.com>
Subject: Re: [RFC 2/4] sched/numa: replace runnable_load_avg by load_avg
Date: Thu, 13 Feb 2020 17:02:21 +0000	[thread overview]
Message-ID: <20200213170220.GA3466@techsingularity.net> (raw)
In-Reply-To: <CAKfTPtD8k-LMaXz_MNmxeW5aXDO4ZZ6j=gwCRTRU89OJ9nUEGw@mail.gmail.com>

On Thu, Feb 13, 2020 at 05:38:31PM +0100, Vincent Guittot wrote:
> > > Your test doesn't explicitly ensure that the 1 condition is met
> > >
> > > That being said, I'm not sure it's really a wrong thing ? I mean
> > > load_balance will probably try to pull back some tasks on src but as
> > > long as it is not a task with dst node as preferred node, it should
> > > not be that harmfull
> >
> > My thinking was that if source has as many or more running tasks than
> > the destination *after* the move that it's not harmful and does not add
> > work for the load balancer.
> 
> load_balancer will see an imbalance but fbq_classify_group/queue
> should be there to prevent from pulling back tasks that are on the
> preferred node but only other tasks
> 

Yes, exactly. Between fbq_classify and migrate_degrades_locality, I'm
expecting that the load balancer will only override NUMA balancing when
there is no better option. When the imbalance check, I want to avoid
the situation where NUMA balancing moves a task for locality, LB pulls
it back for balance, NUMA retries the move etc because it's stupid. The
locality matters but being continually dequeue/enqueue is unhelpful.

While there might be grounds for relaxing the degree an imbalance is
allowed across SD domains, I am avoiding looking in that direction again
until the load balancer and NUMA balancer stop overriding each other for
silly reasons (or the NUMA balancer fighting itself which can happen).

-- 
Mel Gorman
SUSE Labs

  reply	other threads:[~2020-02-13 17:02 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-02-11 17:46 [PATCH 0/4] remove runnable_load_avg and improve group_classify Vincent Guittot
2020-02-11 17:46 ` [PATCH 1/4] sched/fair: reorder enqueue/dequeue_task_fair path Vincent Guittot
2020-02-12 13:20   ` Mel Gorman
2020-02-12 14:47     ` Vincent Guittot
2020-02-12 16:11       ` Mel Gorman
2020-02-11 17:46 ` [RFC 2/4] sched/numa: replace runnable_load_avg by load_avg Vincent Guittot
2020-02-12 13:37   ` Mel Gorman
2020-02-12 15:03     ` Vincent Guittot
2020-02-12 16:04       ` Mel Gorman
2020-02-12 19:49     ` Mel Gorman
2020-02-12 21:29       ` Mel Gorman
2020-02-13  8:05       ` Vincent Guittot
2020-02-13  9:24         ` Mel Gorman
     [not found]         ` <20200213131658.9600-1-hdanton@sina.com>
2020-02-13 13:46           ` Mel Gorman
2020-02-13 15:00             ` Phil Auld
2020-02-13 15:14               ` Mel Gorman
2020-02-13 16:11                 ` Vincent Guittot
2020-02-13 16:34                   ` Mel Gorman
2020-02-13 16:38                     ` Vincent Guittot
2020-02-13 17:02                       ` Mel Gorman [this message]
2020-02-13 17:15                         ` Vincent Guittot
2020-02-11 17:46 ` [RFC 3/4] sched/fair: replace runnable load average by runnable average Vincent Guittot
2020-02-12 14:30   ` Mel Gorman
2020-02-14  7:42     ` Vincent Guittot
2020-02-13 17:36   ` Peter Zijlstra
2020-02-14  7:43     ` Vincent Guittot
2020-02-11 17:46 ` [RFC 4/4] sched/fair: Take into runnable_avg to classify group Vincent Guittot
2020-02-13 18:32   ` Valentin Schneider
2020-02-13 18:37     ` Valentin Schneider
2020-02-14  7:48       ` Vincent Guittot
2020-02-11 21:04 ` [PATCH 0/4] remove runnable_load_avg and improve group_classify Mel Gorman
2020-02-12  8:16   ` Vincent Guittot

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=20200213170220.GA3466@techsingularity.net \
    --to=mgorman@techsingularity.net \
    --cc=bsegall@google.com \
    --cc=dietmar.eggemann@arm.com \
    --cc=hdanton@sina.com \
    --cc=juri.lelli@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=parth@linux.ibm.com \
    --cc=pauld@redhat.com \
    --cc=peterz@infradead.org \
    --cc=rostedt@goodmis.org \
    --cc=valentin.schneider@arm.com \
    --cc=vincent.guittot@linaro.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.