From: David Rientjes <rientjes@google.com> To: Satoru Moriya <satoru.moriya@hds.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>, Andrew Morton <akpm@linux-foundation.org>, Rik van Riel <riel@redhat.com>, Randy Dunlap <rdunlap@xenotime.net>, Satoru Moriya <smoriya@redhat.com>, "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>, "linux-mm@kvack.org" <linux-mm@kvack.org>, "lwoodman@redhat.com" <lwoodman@redhat.com>, Seiji Aguchi <saguchi@redhat.com>, Hugh Dickins <hughd@google.com>, "hannes@cmpxchg.org" <hannes@cmpxchg.org> Subject: RE: [PATCH -v2 -mm] add extra free kbytes tunable Date: Fri, 14 Oct 2011 15:46:36 -0700 (PDT) [thread overview] Message-ID: <alpine.DEB.2.00.1110141536520.21305@chino.kir.corp.google.com> (raw) In-Reply-To: <65795E11DBF1E645A09CEC7EAEE94B9CB4F747AC@USINDEVS02.corp.hds.com> On Fri, 14 Oct 2011, Satoru Moriya wrote: > > Satoru was specifically talking about the VM using free memory for > > pagecache, > > Yes, because we can't stop increasing pagecache and it > occupies RAM where some people want to keep free for bursty > memory requirement. Usually it works fine but sometimes like > my test case doesn't work well. > > > so doing echo echo 1 > /proc/sys/vm/drop_caches can mitigate > > that almost immediately. > > I know it and some admins use that kind of tuning. But is it > proper way? Should we exec the script like above periodically? > I believe that we should use it for debug only. > Agreed, this was in response to the suggestion for adding a mem_shrink() syscall, which would require the same periodic calls or knowledge of the application prior to the bursty memory allocations. I bring up drop_caches just to illustrate that it is effectively the same thing for the entire address space when pressured by pagecache. So I don't think that syscall would actually help for your scenario. > > If there were a change to increase the space significantly between the > > high and min watermark when min_free_kbytes changes, that would fix the > > problem. > > Right. But min_free_kbytes changes both thresholds, foregroud reclaim > and background reclaim. I'd like to configure them separately like > dirty_bytes and dirty_background_bytes for flexibility. > The point I'm trying to make is that if kswapd can be made aware that it was kicked by a rt_task() in the page allocator, the same criteria we use for ALLOC_HARDER today, or a rt_task() subsequently enters the page allocator slowpath while kswapd is running, then not only can we increase the scheduling priority of kswapd but it is also possible to reclaim above the high watermark for an extra bonus. I believe we can find a sane middle ground that requires no userspace tunable where a _single_ realtime application cannot allocate memory faster than kswapd with very high priority and reclaiming above the high watermark, whether that's a factor of 1.25 or not. > > The problem is two-fold: that comes at a penalty for systems > > or workloads that don't need to reclaim the additional memory, and it's > > not clear how much space should exist between those watermarks. > > The required size depends on a system architacture such as kernel, > applications, storage etc. and so admin who care the whole system > should configure it based on tests by his own risk. > Doing that comes at a penalty for other workloads that are running on the same system, which is the problem with a global tunable that doesn't discriminate on an allocator's priority (the min -> high watermarks for reclaim do well except for rt-threads, as evidenced by this thread).
WARNING: multiple messages have this Message-ID (diff)
From: David Rientjes <rientjes@google.com> To: Satoru Moriya <satoru.moriya@hds.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>, Andrew Morton <akpm@linux-foundation.org>, Rik van Riel <riel@redhat.com>, Randy Dunlap <rdunlap@xenotime.net>, Satoru Moriya <smoriya@redhat.com>, "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>, "linux-mm@kvack.org" <linux-mm@kvack.org>, "lwoodman@redhat.com" <lwoodman@redhat.com>, Seiji Aguchi <saguchi@redhat.com>, Hugh Dickins <hughd@google.com>, "hannes@cmpxchg.org" <hannes@cmpxchg.org> Subject: RE: [PATCH -v2 -mm] add extra free kbytes tunable Date: Fri, 14 Oct 2011 15:46:36 -0700 (PDT) [thread overview] Message-ID: <alpine.DEB.2.00.1110141536520.21305@chino.kir.corp.google.com> (raw) In-Reply-To: <65795E11DBF1E645A09CEC7EAEE94B9CB4F747AC@USINDEVS02.corp.hds.com> On Fri, 14 Oct 2011, Satoru Moriya wrote: > > Satoru was specifically talking about the VM using free memory for > > pagecache, > > Yes, because we can't stop increasing pagecache and it > occupies RAM where some people want to keep free for bursty > memory requirement. Usually it works fine but sometimes like > my test case doesn't work well. > > > so doing echo echo 1 > /proc/sys/vm/drop_caches can mitigate > > that almost immediately. > > I know it and some admins use that kind of tuning. But is it > proper way? Should we exec the script like above periodically? > I believe that we should use it for debug only. > Agreed, this was in response to the suggestion for adding a mem_shrink() syscall, which would require the same periodic calls or knowledge of the application prior to the bursty memory allocations. I bring up drop_caches just to illustrate that it is effectively the same thing for the entire address space when pressured by pagecache. So I don't think that syscall would actually help for your scenario. > > If there were a change to increase the space significantly between the > > high and min watermark when min_free_kbytes changes, that would fix the > > problem. > > Right. But min_free_kbytes changes both thresholds, foregroud reclaim > and background reclaim. I'd like to configure them separately like > dirty_bytes and dirty_background_bytes for flexibility. > The point I'm trying to make is that if kswapd can be made aware that it was kicked by a rt_task() in the page allocator, the same criteria we use for ALLOC_HARDER today, or a rt_task() subsequently enters the page allocator slowpath while kswapd is running, then not only can we increase the scheduling priority of kswapd but it is also possible to reclaim above the high watermark for an extra bonus. I believe we can find a sane middle ground that requires no userspace tunable where a _single_ realtime application cannot allocate memory faster than kswapd with very high priority and reclaiming above the high watermark, whether that's a factor of 1.25 or not. > > The problem is two-fold: that comes at a penalty for systems > > or workloads that don't need to reclaim the additional memory, and it's > > not clear how much space should exist between those watermarks. > > The required size depends on a system architacture such as kernel, > applications, storage etc. and so admin who care the whole system > should configure it based on tests by his own risk. > Doing that comes at a penalty for other workloads that are running on the same system, which is the problem with a global tunable that doesn't discriminate on an allocator's priority (the min -> high watermarks for reclaim do well except for rt-threads, as evidenced by this thread). -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2011-10-14 22:46 UTC|newest] Thread overview: 102+ messages / expand[flat|nested] mbox.gz Atom feed top 2011-09-01 14:52 [PATCH -mm] add extra free kbytes tunable Rik van Riel 2011-09-01 14:52 ` Rik van Riel 2011-09-01 17:06 ` Randy Dunlap 2011-09-01 17:06 ` Randy Dunlap 2011-09-01 19:26 ` [PATCH -v2 " Rik van Riel 2011-09-01 19:26 ` Rik van Riel 2011-09-01 21:58 ` Andrew Morton 2011-09-01 21:58 ` Andrew Morton 2011-09-01 22:08 ` David Rientjes 2011-09-01 22:08 ` David Rientjes 2011-09-01 22:16 ` Andrew Morton 2011-09-01 22:16 ` Andrew Morton 2011-09-02 16:31 ` Satoru Moriya 2011-09-02 16:31 ` Satoru Moriya 2011-10-13 7:33 ` Minchan Kim 2011-10-13 7:33 ` Minchan Kim 2011-10-13 8:09 ` KAMEZAWA Hiroyuki 2011-10-13 8:09 ` KAMEZAWA Hiroyuki [not found] ` <E1FA588BC672D846BDBB452FCA1E308C2389B4@USINDEVS02.corp.hds.com> 2011-09-15 3:33 ` Satoru Moriya 2011-09-15 3:33 ` Satoru Moriya 2011-09-01 22:09 ` Andrew Morton 2011-09-01 22:09 ` Andrew Morton 2011-09-02 16:26 ` [PATCH -mm] fixes & cleanups for "add extra free kbytes tunable" Rik van Riel 2011-09-02 16:26 ` Rik van Riel 2011-09-30 21:43 ` [PATCH -v2 -mm] add extra free kbytes tunable Johannes Weiner 2011-09-30 21:43 ` Johannes Weiner 2011-10-08 3:08 ` David Rientjes 2011-10-08 3:08 ` David Rientjes 2011-10-10 22:37 ` Andrew Morton 2011-10-10 22:37 ` Andrew Morton 2011-10-11 19:32 ` Satoru Moriya 2011-10-11 19:32 ` Satoru Moriya 2011-10-11 19:54 ` Andrew Morton 2011-10-11 19:54 ` Andrew Morton 2011-10-11 20:23 ` Satoru Moriya 2011-10-11 20:23 ` Satoru Moriya 2011-10-11 20:54 ` Andrew Morton 2011-10-11 20:54 ` Andrew Morton 2011-10-12 13:09 ` Rik van Riel 2011-10-12 13:09 ` Rik van Riel 2011-10-12 19:20 ` Andrew Morton 2011-10-12 19:20 ` Andrew Morton 2011-10-12 19:58 ` Rik van Riel 2011-10-12 19:58 ` Rik van Riel 2011-10-12 20:26 ` David Rientjes 2011-10-12 20:26 ` David Rientjes 2011-10-21 23:48 ` Satoru Moriya 2011-10-21 23:48 ` Satoru Moriya 2011-10-23 21:22 ` David Rientjes 2011-10-23 21:22 ` David Rientjes 2011-10-25 2:04 ` Satoru Moriya 2011-10-25 2:04 ` Satoru Moriya 2011-10-25 21:50 ` David Rientjes 2011-10-25 21:50 ` David Rientjes 2011-10-26 18:59 ` Satoru Moriya 2011-10-26 18:59 ` Satoru Moriya 2011-10-12 21:08 ` Satoru Moriya 2011-10-12 21:08 ` Satoru Moriya 2011-10-12 22:41 ` David Rientjes 2011-10-12 22:41 ` David Rientjes 2011-10-12 23:52 ` Satoru Moriya 2011-10-12 23:52 ` Satoru Moriya 2011-10-13 0:01 ` David Rientjes 2011-10-13 0:01 ` David Rientjes 2011-10-13 5:35 ` KAMEZAWA Hiroyuki 2011-10-13 5:35 ` KAMEZAWA Hiroyuki 2011-10-13 20:55 ` David Rientjes 2011-10-13 20:55 ` David Rientjes 2011-10-14 22:16 ` Satoru Moriya 2011-10-14 22:16 ` Satoru Moriya 2011-10-14 22:46 ` David Rientjes [this message] 2011-10-14 22:46 ` David Rientjes 2011-10-14 5:32 ` Satoru Moriya 2011-10-14 5:32 ` Satoru Moriya 2011-10-14 5:06 ` Satoru Moriya 2011-10-14 5:06 ` Satoru Moriya 2011-10-11 23:22 ` David Rientjes 2011-10-11 23:22 ` David Rientjes 2011-10-13 16:54 ` Satoru Moriya 2011-10-13 16:54 ` Satoru Moriya 2011-10-13 20:48 ` David Rientjes 2011-10-13 20:48 ` David Rientjes 2011-10-13 21:11 ` Rik van Riel 2011-10-13 21:11 ` Rik van Riel 2011-10-13 22:02 ` David Rientjes 2011-10-13 22:02 ` David Rientjes 2011-10-11 19:20 ` Satoru Moriya 2011-10-11 19:20 ` Satoru Moriya 2011-10-11 21:04 ` David Rientjes 2011-10-11 21:04 ` David Rientjes 2011-10-12 13:13 ` Rik van Riel 2011-10-12 13:13 ` Rik van Riel 2011-10-12 20:21 ` David Rientjes 2011-10-12 20:21 ` David Rientjes 2011-10-13 4:13 ` Rik van Riel 2011-10-13 4:13 ` Rik van Riel 2011-10-13 5:22 ` David Rientjes 2011-10-13 5:22 ` David Rientjes 2011-10-22 0:11 ` Satoru Moriya 2011-10-22 0:11 ` Satoru Moriya 2011-09-09 23:01 Satoru Moriya 2011-09-09 23:01 ` Satoru Moriya
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=alpine.DEB.2.00.1110141536520.21305@chino.kir.corp.google.com \ --to=rientjes@google.com \ --cc=akpm@linux-foundation.org \ --cc=hannes@cmpxchg.org \ --cc=hughd@google.com \ --cc=kamezawa.hiroyu@jp.fujitsu.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=lwoodman@redhat.com \ --cc=rdunlap@xenotime.net \ --cc=riel@redhat.com \ --cc=saguchi@redhat.com \ --cc=satoru.moriya@hds.com \ --cc=smoriya@redhat.com \ /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: linkBe 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.