From: Srivatsa Vaddagiri <vatsa@in.ibm.com>
To: "Paul Menage" <menage@google.com>
Cc: sekharan@us.ibm.com, ckrm-tech@lists.sourceforge.net,
linux-kernel@vger.kernel.org, xemul@sw.ru, rohitseth@google.com,
pj@sgi.com, "Eric W. Biederman" <ebiederm@xmission.com>,
mbligh@google.com, winget@google.com, containers@lists.osdl.org,
"Serge E. Hallyn" <serue@us.ibm.com>,
dev@sw.ru, devel@openvz.org
Subject: Re: [ckrm-tech] [PATCH 7/7] containers (V7): Container interface to nsproxy subsystem
Date: Thu, 5 Apr 2007 14:19:20 +0530 [thread overview]
Message-ID: <20070405084920.GB20356@in.ibm.com> (raw)
In-Reply-To: <6599ad830704042348q5675c01ep92679fb3275bafac@mail.gmail.com>
On Wed, Apr 04, 2007 at 11:48:57PM -0700, Paul Menage wrote:
> >rcfs_task_count will essentially return number of tasks pointing to A1
> >thr' their nsproxy->ctlr_data[BC_ID].
>
> One small issue with the (last posted) version of your patch is that
> it doesn't take into account the refcounts from the directories
> themselves
You mean dentry->d_fsdata pointing to nsproxy should take a ref count on
nsproxy? afaics it is not needed as long as you first drop the dentry
before freeing associated nsproxy.
> - I think you probably need to subtract one for each active
> subsystem.
I don't understand this.
> I don't think that's a reasonable assumption. A task can have
> thousands of file handles open - having to scan and move every file
> that the task has open would make a move operation incredibly
> expensive.
>
> Additionally, tasks can share many of those file handles
> with other tasks. So what happens if one task that has a file open
> moves out of the container, but another stays behind? It's cleaner and
> more efficient, and conceptually desirable, IMO, just to keep the file
> associated with the container.
I don't have a authoritative view here on whether open file count should
be migrated or not, but from a layman perspective consider this:
- Task T1 is in Container C1, whose max open files can be 100
- T1 opens all of those 100 files
- T1 migrates to Container C2, but its open file count is not
migrated
- T2 is migrated to container C1 and tries opening a file but is
denied. T2 looks for "who is in my container who has opened all
files" and doesn't find anyone.
Isn't that a bit abnormal from an end-user pov?
> >Why refcount 3? I can only be 1 (from T) ..
>
> Plus the refcounts from the two filesystem roots.
Filesystem root dentry's are special case. They will point to
init_nsproxy which is never deleted and hence they need not add
additional ref counts.
For other directories created, say H1/foo, foo's dentry will point to N1
but need not take additional refcount. N1 won't be deleted w/o dropping
foo's dentry first. I think this is very similar to cpuset case, where
dentry->d_fsdata = cs doesnt take additional ref counts on cpuset.
> >The object was created by the task, so I would expect it should get
> >migrated too to the new task's context (which should be true in case of
> >f_bc atleast?). Can you give a practical example where you want to
> >migrate the task and not the object it created?
>
> I gave one above, for files; others could include pages (do you want
> to have to migrate every page when a task switches container? what
> about shared pages?)
>
> Obviously this fundamental difference of opinion means that we're
> going to end up disagreeing on whether the scenario I presented is a
> problem or not ...
Again I am not a VM expert to say whether pages should get migrated or
not. But coming to the impact of this discussion on xxx_rmdir() ..
> The problem with that is that (given the assumption that some
> subsystems might not want to migrate objects) you can then end up with
> a subsystem state object that has refcounts on it from active objects
> like files, but which is unreachable via any container filesystem
> mechanism. Better IMO to be able to fail the rmdir() in that situation
> so that the subsystem object remains accessible (so you can see where
> the resources are being used up).
I agree we shouldn't delete a dir going by just the task count. How abt
a (optional) ->can_destroy callback which will return -EBUSY if additional
non-task objects are pointing to a subsyste's resource object?
--
Regards,
vatsa
next prev parent reply other threads:[~2007-04-05 8:41 UTC|newest]
Thread overview: 132+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-02-12 8:15 [PATCH 0/7] containers (V7): Generic Process Containers menage
2007-02-12 8:15 ` [PATCH 1/7] containers (V7): Generic container system abstracted from cpusets code menage
2007-02-12 12:33 ` Srivatsa Vaddagiri
2007-02-12 19:26 ` Paul Menage
2007-02-12 19:46 ` Paul Menage
2007-02-13 5:48 ` Srivatsa Vaddagiri
2007-02-13 8:16 ` [ckrm-tech] " Srivatsa Vaddagiri
2007-03-07 12:21 ` Srivatsa Vaddagiri
2007-03-07 14:06 ` [ckrm-tech] " Srivatsa Vaddagiri
2007-03-07 20:50 ` Paul Menage
2007-03-08 10:38 ` [ckrm-tech] " Srivatsa Vaddagiri
2007-03-08 10:40 ` Paul Menage
2007-03-11 19:38 ` Paul Jackson
2007-03-12 14:19 ` Srivatsa Vaddagiri
2007-03-22 9:56 ` Srivatsa Vaddagiri
2007-03-22 10:20 ` Srivatsa Vaddagiri
2007-03-24 15:05 ` Srivatsa Vaddagiri
2007-03-24 19:25 ` Paul Jackson
2007-03-25 0:45 ` Srivatsa Vaddagiri
2007-03-25 1:41 ` Paul Jackson
2007-03-25 2:28 ` Srivatsa Vaddagiri
2007-03-25 4:16 ` Srivatsa Vaddagiri
2007-03-25 5:43 ` Paul Jackson
2007-03-25 8:21 ` Srivatsa Vaddagiri
2007-03-25 4:45 ` Paul Jackson
2007-03-25 5:05 ` Srivatsa Vaddagiri
2007-03-25 4:59 ` Paul Jackson
2007-02-12 8:15 ` [PATCH 2/7] containers (V7): Cpusets hooked into containers menage
2007-02-15 20:35 ` Serge E. Hallyn
2007-02-15 20:49 ` Paul Menage
2007-03-07 14:34 ` Srivatsa Vaddagiri
2007-03-07 16:01 ` Paul Menage
2007-03-07 16:31 ` [ckrm-tech] " Srivatsa Vaddagiri
2007-03-07 16:31 ` Paul Menage
2007-03-07 14:52 ` Srivatsa Vaddagiri
2007-03-07 16:12 ` Paul Menage
2007-02-12 8:15 ` [PATCH 4/7] containers (V7): Simple CPU accounting container subsystem menage
2007-02-12 8:15 ` [PATCH 5/7] containers (V7): Resource Groups over generic containers menage
2007-02-12 8:15 ` [PATCH 6/7] containers (V7): BeanCounters over generic process containers menage
2007-02-12 15:34 ` Srivatsa Vaddagiri
2007-02-12 18:49 ` Paul Menage
2007-02-13 8:52 ` Pavel Emelianov
2007-02-13 9:03 ` Paul Menage
2007-02-13 9:18 ` Pavel Emelianov
2007-02-13 9:37 ` Paul Menage
2007-02-13 9:49 ` Pavel Emelianov
2007-02-12 8:15 ` [PATCH 7/7] containers (V7): Container interface to nsproxy subsystem menage
2007-03-24 5:05 ` [ckrm-tech] " Srivatsa Vaddagiri
2007-03-24 16:23 ` Srivatsa Vaddagiri
2007-03-26 21:57 ` Serge E. Hallyn
2007-03-28 14:55 ` Srivatsa Vaddagiri
2007-03-28 15:26 ` Serge E. Hallyn
2007-03-26 21:55 ` Serge E. Hallyn
2007-03-31 2:47 ` Srivatsa Vaddagiri
2007-04-02 14:09 ` Serge E. Hallyn
2007-04-02 14:27 ` Srivatsa Vaddagiri
2007-04-02 18:02 ` Eric W. Biederman
2007-04-03 14:16 ` Srivatsa Vaddagiri
2007-04-03 15:32 ` Serge E. Hallyn
2007-04-03 15:45 ` Paul Menage
2007-04-03 15:54 ` Serge E. Hallyn
2007-04-03 16:16 ` Srivatsa Vaddagiri
2007-04-03 16:26 ` Kirill Korotaev
2007-04-03 16:46 ` Srivatsa Vaddagiri
2007-04-03 16:52 ` Paul Menage
2007-04-03 17:11 ` Srivatsa Vaddagiri
2007-04-03 17:10 ` Paul Menage
2007-04-03 17:30 ` Srivatsa Vaddagiri
2007-04-03 17:30 ` Paul Menage
2007-04-03 17:51 ` Srivatsa Vaddagiri
2007-04-03 17:49 ` Paul Menage
2007-04-04 3:07 ` Srivatsa Vaddagiri
2007-04-04 3:44 ` Paul Jackson
2007-04-04 4:04 ` Paul Menage
2007-04-04 5:15 ` Srivatsa Vaddagiri
2007-04-04 7:00 ` Paul Menage
2007-04-04 17:26 ` Srivatsa Vaddagiri
2007-04-04 17:42 ` Srivatsa Vaddagiri
2007-04-04 18:57 ` Paul Menage
2007-04-04 23:02 ` Eric W. Biederman
2007-04-05 1:35 ` Paul Menage
2007-04-05 1:37 ` Paul Menage
2007-04-05 16:57 ` Srivatsa Vaddagiri
2007-04-05 17:14 ` Srivatsa Vaddagiri
2007-04-06 21:54 ` Paul Menage
2007-04-05 2:57 ` Paul Menage
2007-04-05 6:39 ` Srivatsa Vaddagiri
2007-04-05 6:46 ` Srivatsa Vaddagiri
2007-04-05 6:48 ` Paul Menage
2007-04-05 8:49 ` Srivatsa Vaddagiri [this message]
2007-04-05 9:29 ` Paul Menage
2007-04-05 12:43 ` Srivatsa Vaddagiri
2007-04-05 14:13 ` Srivatsa Vaddagiri
2007-04-05 14:13 ` Paul Menage
2007-04-05 14:46 ` Srivatsa Vaddagiri
2007-04-03 17:34 ` Srivatsa Vaddagiri
2007-04-03 17:29 ` Paul Menage
2007-04-03 16:10 ` Srivatsa Vaddagiri
2007-04-03 15:41 ` Serge E. Hallyn
2007-02-12 9:18 ` [PATCH 0/7] containers (V7): Generic Process Containers Paul Jackson
2007-02-12 9:32 ` Paul Menage
2007-02-12 9:52 ` Paul Jackson
[not found] ` <20070212085104.485337000@menage.corp.google.com>
2007-02-12 15:27 ` [PATCH 3/7] containers (V7): Add generic multi-subsystem API to containers Srivatsa Vaddagiri
2007-02-12 18:40 ` Paul Menage
2007-02-13 13:19 ` Srivatsa Vaddagiri
2007-02-15 1:17 ` Paul Menage
2007-02-12 15:39 ` Serge E. Hallyn
2007-02-12 15:56 ` Cedric Le Goater
2007-02-12 18:31 ` Paul Menage
2007-02-14 8:49 ` Balbir Singh
2007-03-08 17:52 ` Srivatsa Vaddagiri
2007-03-24 12:51 ` [ckrm-tech] " Srivatsa Vaddagiri
2007-02-12 22:38 ` [PATCH 0/7] containers (V7): Generic Process Containers Sam Vilain
2007-02-12 22:47 ` Serge E. Hallyn
2007-02-12 23:18 ` Paul Menage
2007-02-12 23:15 ` Paul Menage
2007-02-13 0:30 ` Sam Vilain
2007-02-13 0:42 ` [ckrm-tech] " Paul Menage
2007-02-13 1:13 ` Sam Vilain
2007-02-13 1:47 ` Paul Menage
2007-02-20 17:34 ` Eric W. Biederman
2007-02-20 17:55 ` Paul Menage
2007-02-20 19:29 ` Eric W. Biederman
2007-02-20 22:47 ` Paul Menage
2007-02-20 23:08 ` Sam Vilain
2007-02-20 23:36 ` Paul Menage
2007-02-20 23:32 ` Serge E. Hallyn
2007-02-20 21:58 ` Sam Vilain
2007-02-20 22:19 ` Paul Menage
2007-02-20 22:58 ` Sam Vilain
2007-02-20 23:28 ` Paul Menage
2007-02-20 23:37 ` Serge E. Hallyn
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=20070405084920.GB20356@in.ibm.com \
--to=vatsa@in.ibm.com \
--cc=ckrm-tech@lists.sourceforge.net \
--cc=containers@lists.osdl.org \
--cc=dev@sw.ru \
--cc=devel@openvz.org \
--cc=ebiederm@xmission.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mbligh@google.com \
--cc=menage@google.com \
--cc=pj@sgi.com \
--cc=rohitseth@google.com \
--cc=sekharan@us.ibm.com \
--cc=serue@us.ibm.com \
--cc=winget@google.com \
--cc=xemul@sw.ru \
/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).