public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Chandra Seetharaman <sekharan@us.ibm.com>
To: Kirill Korotaev <dev@sw.ru>
Cc: Andrew Morton <akpm@osdl.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Alan Cox <alan@lxorguk.ukuu.org.uk>, Ingo Molnar <mingo@elte.hu>,
	Christoph Hellwig <hch@infradead.org>,
	Pavel Emelianov <xemul@openvz.org>, Andrey Savochkin <saw@sw.ru>,
	devel@openvz.org, Rik van Riel <riel@redhat.com>,
	hugh@veritas.com, ckrm-tech@lists.sourceforge.net,
	Andi Kleen <ak@suse.de>
Subject: Re: [ckrm-tech] [RFC][PATCH] UBC: user resource beancounters
Date: Fri, 18 Aug 2006 12:39:52 -0700	[thread overview]
Message-ID: <1155929992.26155.60.camel@linuxchandra> (raw)
In-Reply-To: <44E33893.6020700@sw.ru>

Kirill,

Here are some concerns I have (as of now) w.r.t using UBC for resource
management (in the context of resource groups).

- guarantee support is missing. I do not see any code to provide the
  minimum amount of resource a group can get. It is important for 
  providing QoS. (In a different email you did mention guarantee, i am
  referring it here for completeness).
- Creation of a UBC and assignment of task to a UBC always happen in
  the context of the task that is affected. I can understand it works in
  OpenVZ environment, but IMO has issues if one wants it to be used for 
  basic resource management
   - application needs to be changed to use this feature.
   - System administrator does not have the control to assign tasks to a
     UBC. Application does by itself.
   - Assignment of task to a UBC need to be transparent to the 
     application.
- UBC is deleted when the last task (in that UBC) exits. For resource
  management purposes, UBC should be deleted only when the administrator
  deletes it.
- No ability to set resource specific configuration information.
- No ability to maintain resource specific data in the controller.
- No ability to get the list of tasks belonging to a UBC.
- Doesn't inform the resource controllers when limits(shares) change.
- Doesn't inform the resource controllers when a task's UBC has changed.
- Doesn't recalculate the resource usage when a task's UBC has changed. 
  i.e doesn't uncharge the old UBC and charge new UBC.
- For a system administrator name for identification of a UBC is 
  better than a number (uid).

regards,

chandra

On Wed, 2006-08-16 at 19:24 +0400, Kirill Korotaev wrote:
> The following patch set presents base of
> User Resource Beancounters (UBC).
> UBC allows to account and control consumption
> of kernel resources used by group of processes.
> 
> The full UBC patch set allows to control:
> - kernel memory. All the kernel objects allocatable
>   on user demand should be accounted and limited
>   for DoS protection.
>   E.g. page tables, task structs, vmas etc.
> 
> - virtual memory pages. UBC allows to
>   limit a container to some amount of memory and
>   introduces 2-level OOM killer taking into account
>   container's consumption.
>   pages shared between containers are correctly
>   charged as fractions (tunable).
> 
> - network buffers. These includes TCP/IP rcv/snd
>   buffers, dgram snd buffers, unix, netlinks and
>   other buffers.
> 
> - minor resources accounted/limited by number:
>   tasks, files, flocks, ptys, siginfo, pinned dcache
>   mem, sockets, iptentries (for containers with
>   virtualized networking)
> 
> As the first step we want to propose for discussion
> the most complicated parts of resource management:
> kernel memory and virtual memory.
> The patch set to be sent provides core for UBC and
> management of kernel memory only. Virtual memory
> management will be sent in a couple of days.
> 
> The patches in these series are:
> diff-ubc-kconfig.patch:
>     Adds kernel/ub/Kconfig file with UBC options and
>     includes it into arch Kconfigs
> 
> diff-ubc-core.patch:
>     Contains core functionality and interfaces of UBC:
>     find/create beancounter, initialization,
>     charge/uncharge of resource, core objects' declarations.
> 
> diff-ubc-task.patch:
>     Contains code responsible for setting UB on task,
>     it's inheriting and setting host context in interrupts.
> 
>     Task contains three beancounters:
>     1. exec_ub  - current context. all resources are charged
>                   to this beancounter.
>     2. task_ub  - beancounter to which task_struct is charged
>                   itself.
>     3. fork_sub - beancounter which is inherited by
>                   task's children on fork
> 
> diff-ubc-syscalls.patch:
>     Patch adds system calls for UB management:
>     1. sys_getluid    - get current UB id
>     2. sys_setluid    - changes exec_ and fork_ UBs on current
>     3. sys_setublimit - set limits for resources consumtions
> 
> diff-ubc-kmem-core.patch:
>     Introduces UB_KMEMSIZE resource which accounts kernel
>     objects allocated by task's request.
> 
>     Objects are accounted via struct page and slab objects.
>     For the latter ones each slab contains a set of pointers
>     corresponding object is charged to.
> 
>     Allocation charge rules:
>     1. Pages - if allocation is performed with __GFP_UBC flag - page
>        is charged to current's exec_ub.
>     2. Slabs - kmem_cache may be created with SLAB_UBC flag - in this
>        case each allocation is charged. Caches used by kmalloc are
>        created with SLAB_UBC | SLAB_UBC_NOCHARGE flags. In this case
>        only __GFP_UBC allocations are charged.
> 
> diff-ubc-kmem-charge.patch:
>     Adds SLAB_UBC and __GFP_UBC flags in appropriate places
>     to cause charging/limiting of specified resources.
> 
> diff-ubc-proc.patch:
>     Adds two proc entries user_beancounters and user_beancounters_sub
>     allowing to see current state (usage/limits/fails for each UB).
>     Implemented via seq files.
> 
> Patch set is applicable to 2.6.18-rc4-mm1
> 
> Thanks,
> Kirill
> 
> 
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> ckrm-tech mailing list
> https://lists.sourceforge.net/lists/listinfo/ckrm-tech
-- 

----------------------------------------------------------------------
    Chandra Seetharaman               | Be careful what you choose....
              - sekharan@us.ibm.com   |      .......you may get it.
----------------------------------------------------------------------



  parent reply	other threads:[~2006-08-18 19:39 UTC|newest]

Thread overview: 209+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-08-16 15:24 [RFC][PATCH] UBC: user resource beancounters Kirill Korotaev
2006-08-16 15:35 ` [RFC][PATCH 1/7] UBC: kconfig Kirill Korotaev
2006-08-18 19:57   ` [ckrm-tech] " Chandra Seetharaman
2006-08-18 21:14   ` Adrian Bunk
2006-08-25 15:12   ` Pavel Machek
2006-08-16 15:37 ` [RFC][PATCH 2/7] UBC: core (structures, API) Kirill Korotaev
2006-08-16 16:58   ` Alan Cox
2006-08-17 11:42     ` Kirill Korotaev
2006-08-16 17:15   ` Greg KH
2006-08-17 11:45     ` Kirill Korotaev
2006-08-17 12:14       ` Greg KH
2006-08-17 14:32       ` [ckrm-tech] " Dave Hansen
2006-08-18 12:36         ` Kirill Korotaev
2006-08-16 18:11   ` Rohit Seth
2006-08-16 18:18     ` Andrew Morton
2006-08-17 11:54       ` Kirill Korotaev
2006-08-17 11:53     ` Kirill Korotaev
2006-08-17 16:55       ` Rohit Seth
2006-08-18 11:14         ` Kirill Korotaev
2006-08-18 17:51           ` Rohit Seth
2006-08-18  5:31       ` Andrew Morton
2006-08-18  7:35         ` [PATCH " Andrey Savochkin
2006-08-18  8:26           ` [ckrm-tech] " Matt Helsley
2006-08-18 11:52             ` Kirill Korotaev
2006-08-18 15:59         ` [RFC][PATCH " Alan Cox
2006-08-17 11:09   ` [ckrm-tech] " Srivatsa Vaddagiri
2006-08-17 14:02     ` Kirill Korotaev
2006-08-17 18:59       ` Chandra Seetharaman
2006-08-18  1:58   ` Matt Helsley
2006-08-18 11:36     ` Kirill Korotaev
2006-08-19  2:38       ` Matt Helsley
2006-08-21 11:02         ` Kirill Korotaev
2006-08-22 12:23           ` Srivatsa Vaddagiri
2006-08-22 12:46             ` Kirill Korotaev
2006-08-22 14:38               ` Srivatsa Vaddagiri
2006-08-21 17:35         ` Chandra Seetharaman
2006-08-20  4:58   ` Balbir Singh
2006-08-20  5:01   ` Balbir Singh
2006-08-16 15:38 ` [RFC][PATCH 3/7] UBC: ub context and inheritance Kirill Korotaev
2006-08-16 16:51   ` Alan Cox
2006-08-17 11:09   ` [ckrm-tech] " Srivatsa Vaddagiri
2006-08-17 13:21     ` [Devel] " Pavel V. Emelianov
2006-08-18  2:42   ` Matt Helsley
2006-08-18  9:23     ` Kirill Korotaev
2006-08-19  2:19       ` Matt Helsley
2006-08-18 20:03   ` Chandra Seetharaman
2006-08-21 10:32     ` Kirill Korotaev
2006-08-21 20:48       ` Chandra Seetharaman
2006-08-16 15:39 ` [RFC][PATCH 4/7] UBC: syscalls (user interface) Kirill Korotaev
2006-08-16 16:52   ` Alan Cox
2006-08-16 17:17   ` Greg KH
2006-08-17 12:02     ` Kirill Korotaev
2006-08-16 18:17   ` Rohit Seth
2006-08-16 19:04     ` Alan Cox
2006-08-16 19:22       ` Rohit Seth
2006-08-17 12:13         ` Kirill Korotaev
2006-08-17 15:40           ` Andrew Morton
2006-08-18  8:08             ` [PATCH " Andrey Savochkin
2006-08-18 14:45               ` [ckrm-tech] " Dave Hansen
2006-08-18 16:42                 ` Andrew Morton
2006-08-18 17:29                   ` Dave Hansen
2006-08-18 17:38                     ` Andrew Morton
2006-08-18 17:59                   ` Rohit Seth
2006-08-18 18:18                     ` Andrew Morton
2006-08-21  2:42                     ` Magnus Damm
2006-08-18 18:09                   ` Paul Jackson
2006-08-18 18:17                   ` Chandra Seetharaman
2006-08-18 18:27                     ` Chandra Seetharaman
2006-08-18 18:56                     ` Paul Jackson
2006-08-18 19:16                       ` Chris Friesen
2006-08-18 21:19                         ` Paul Jackson
2006-08-18 19:48                       ` Chandra Seetharaman
2006-08-18 21:16                         ` Paul Jackson
2006-08-21  2:38                   ` Magnus Damm
2006-08-21  7:48                     ` Andi Kleen
2006-08-21  8:42                       ` Magnus Damm
2006-08-21  9:03                         ` Andi Kleen
2006-08-21  9:18                           ` Magnus Damm
2006-08-21 13:35                   ` Kirill Korotaev
2006-08-21 17:51                     ` Paul Jackson
2006-08-22  8:52                       ` Kirill Korotaev
2006-08-21  2:47                 ` Magnus Damm
2006-08-22  1:16                   ` Rohit Seth
2006-08-22  3:58                     ` Magnus Damm
2006-08-22 18:34                       ` Chandra Seetharaman
2006-08-24  1:20                       ` Rohit Seth
2006-08-18 11:05             ` [RFC][PATCH " Kirill Korotaev
2006-08-17 17:08           ` Rohit Seth
2006-08-17 12:04     ` Kirill Korotaev
2006-08-17 17:05       ` Rohit Seth
2006-08-17 11:09   ` [ckrm-tech] " Srivatsa Vaddagiri
2006-08-17 14:04     ` Kirill Korotaev
2006-08-17 16:19       ` Srivatsa Vaddagiri
2006-08-18  2:31   ` Matt Helsley
2006-08-18 11:45     ` Kirill Korotaev
2006-08-19  2:43       ` Matt Helsley
2006-08-18 11:40   ` Arnd Bergmann
2006-08-18 20:13   ` [ckrm-tech] " Chandra Seetharaman
2006-08-16 15:40 ` [RFC][PATCH 5/7] UBC: kernel memory accounting (core) Kirill Korotaev
2006-08-16 16:56   ` Alan Cox
2006-08-17 13:47     ` Kirill Korotaev
2006-08-16 18:24   ` Rohit Seth
2006-08-17 13:27     ` Kirill Korotaev
2006-08-17 14:38       ` [ckrm-tech] " Dave Hansen
2006-08-18  9:31         ` Kirill Korotaev
2006-08-18 14:58           ` Dave Hansen
2006-08-21 10:40             ` Kirill Korotaev
2006-08-21 15:10               ` Dave Hansen
2006-08-18 15:06           ` Dave Hansen
2006-08-21 12:38             ` Kirill Korotaev
2006-08-17 17:02       ` Rohit Seth
2006-08-18  9:38         ` Kirill Korotaev
2006-08-18 16:55           ` Rohit Seth
2006-08-21 10:43             ` Kirill Korotaev
2006-08-22  1:23               ` Rohit Seth
2006-08-16 18:47   ` Dave Hansen
2006-08-16 19:15     ` Rohit Seth
2006-08-16 19:59       ` [ckrm-tech] " Dave Hansen
2006-08-17  0:24         ` Alan Cox
2006-08-17 14:26           ` Dave Hansen
2006-08-17 15:01             ` Alan Cox
2006-08-17 16:04               ` Andi Kleen
2006-08-18 10:54                 ` Kirill Korotaev
2006-08-17 16:37               ` Dave Hansen
2006-08-17 15:19             ` Rik van Riel
2006-08-17 17:28               ` Rohit Seth
2006-08-17 18:43                 ` Andi Kleen
2006-08-17 17:49                   ` Dave Hansen
2006-08-18  8:29               ` Kirill Korotaev
2006-08-18 17:06                 ` Rohit Seth
2006-08-17 17:16             ` Rohit Seth
2006-08-17 17:23               ` Dave Hansen
2006-08-17 17:36                 ` Rohit Seth
2006-08-17 17:53                   ` Dave Hansen
2006-08-18  8:54                   ` Kirill Korotaev
2006-08-18 14:52                     ` Dave Hansen
2006-08-18 17:38                     ` Rohit Seth
2006-08-21 11:29                       ` Kirill Korotaev
2006-08-22  1:48                         ` Rohit Seth
2006-08-22  7:43                           ` Pavel V. Emelianov
2006-08-18  8:51               ` Kirill Korotaev
2006-08-18  8:52               ` Kirill Korotaev
2006-08-18 14:59                 ` Alan Cox
2006-08-18 19:32                   ` Dave Hansen
2006-08-18 20:52                     ` Alan Cox
2006-08-21  9:44                       ` Kirill Korotaev
2006-08-17 16:42           ` Rohit Seth
2006-08-17 16:31         ` Rohit Seth
2006-08-17  0:22       ` Alan Cox
2006-08-17 16:36         ` Rohit Seth
2006-08-18  8:44           ` [ckrm-tech] " Kirill Korotaev
2006-08-17 13:35       ` Kirill Korotaev
2006-08-17 17:13         ` Rohit Seth
2006-08-18  8:49           ` [ckrm-tech] " Kirill Korotaev
2006-08-17 13:31     ` Kirill Korotaev
2006-08-17 14:36       ` Dave Hansen
2006-08-18  8:12         ` [ckrm-tech] " Kirill Korotaev
2006-08-18 14:43           ` Dave Hansen
2006-08-21  8:57             ` Kirill Korotaev
2006-08-18 20:26   ` Chandra Seetharaman
2006-08-21 10:51     ` Kirill Korotaev
2006-08-21 20:55       ` Chandra Seetharaman
2006-08-16 15:42 ` [RFC][PATCH 6/7] UBC: kernel memory acconting (mark objects) Kirill Korotaev
2006-08-16 16:57   ` Alan Cox
2006-08-16 15:44 ` [RFC][PATCH 7/7] UBC: proc interface Kirill Korotaev
2006-08-16 17:13   ` Greg KH
2006-08-17 13:43     ` Kirill Korotaev
2006-08-17 15:40       ` Greg KH
2006-08-17 16:12         ` [Devel] " Kir Kolyshkin
2006-08-16 18:53 ` [RFC][PATCH] UBC: user resource beancounters Rohit Seth
2006-08-16 19:26   ` Alan Cox
2006-08-17  0:15 ` [ckrm-tech] " Chandra Seetharaman
2006-08-17 11:02 ` Srivatsa Vaddagiri
2006-08-17 13:55   ` Kirill Korotaev
2006-08-17 19:55     ` Chandra Seetharaman
2006-08-18 10:36       ` Kirill Korotaev
2006-08-18 18:53         ` Chandra Seetharaman
2006-08-18 22:55           ` Matt Helsley
2006-08-21 10:55           ` Kirill Korotaev
2006-08-21 21:04             ` Chandra Seetharaman
2006-08-18 19:39 ` Chandra Seetharaman [this message]
2006-08-21 13:24   ` Kirill Korotaev
2006-08-21 21:45     ` Chandra Seetharaman
2006-08-21 22:20       ` Alan Cox
2006-08-21 22:44         ` Chandra Seetharaman
2006-08-22  1:45       ` Rohit Seth
2006-08-22 10:02         ` Alan Cox
2006-08-22  9:57           ` Arjan van de Ven
2006-08-22 11:15             ` Alan Cox
2006-08-24  1:31           ` Rohit Seth
2006-08-22 18:55         ` Chandra Seetharaman
2006-08-24  1:44           ` Rohit Seth
2006-08-24  2:04             ` Chandra Seetharaman
2006-08-24 11:10               ` Alan Cox
2006-08-24 23:48                 ` Chandra Seetharaman
2006-08-24 23:55                   ` Kyle Moffett
2006-08-25 18:21                     ` Chandra Seetharaman
2006-08-25 20:46                       ` Alan Cox
2006-08-25 21:37                         ` Chandra Seetharaman
2006-08-25 22:51                           ` Alan Cox
2006-08-25 22:59                             ` Chandra Seetharaman
2006-08-24 17:27               ` Rohit Seth
2006-08-24 23:52                 ` Chandra Seetharaman
2006-08-25 11:12                   ` Kirill Korotaev
2006-08-25 18:47                     ` Chandra Seetharaman
2006-08-25 20:52                       ` Alan Cox
2006-08-25 22:23                         ` Chandra Seetharaman
2006-08-25 23:12                           ` Alan Cox
2006-08-25 23:00                             ` Chandra Seetharaman

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=1155929992.26155.60.camel@linuxchandra \
    --to=sekharan@us.ibm.com \
    --cc=ak@suse.de \
    --cc=akpm@osdl.org \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=ckrm-tech@lists.sourceforge.net \
    --cc=dev@sw.ru \
    --cc=devel@openvz.org \
    --cc=hch@infradead.org \
    --cc=hugh@veritas.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=riel@redhat.com \
    --cc=saw@sw.ru \
    --cc=xemul@openvz.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