Guys, JFYI, we are getting rid of LOG_NONL and indent in logger as non
multi-thread safe (we are moving towards multi threaded eventloop):
https://git.acronis.com/projects/STRG/repos/pstorage-core/pull-requests/765/diff
On 5 Dec 2018, at 12:29, Pavel Butsykin
mailto:pbutsy...@virtuozzo.com>>
Tejun,
Sorry for getting into the middle of the discussion, but...
Can you imagine how many userland APIs are needed to make userspace C/R?
Do you really want APIs in user-space which allow to:
- send signals with siginfo attached (kill() doesn't work...)
- read inotify configuration
- insert
Arn wrote:
Hi,
I need to know the following for some experiments I'm doing as part of
a school project:
Does OpenVZ support host swapping of guest pages ? That is : if a host
experiences memory pressure, can it swap out some of a guest's (guest
vm) pages and use them ? I assume that if
Arn wrote:
On Sun, Jul 6, 2008 at 2:19 PM, Kirill Korotaev [EMAIL PROTECTED] wrote:
Does OpenVZ support host swapping of guest pages ? That is : if a host
experiences memory pressure, can it swap out some of a guest's (guest
vm) pages and use them ? I assume that if this is possible
This kind of error says that Oracle couldn't find the libraries.
Check your settings and that library is foundable in $LD_LIBRARY path.
Also google. I remember I had similar problems due to Oracle configuration
issues.
杜雨阳 wrote:
HI,
Here is the thing. Oracle 10g was first installed without
we will prefer
simply to have
a compatibility patch for specific kernel versions in our kernel tree (not in
mainstream).
Definitely, other ideas/opinions are welcome.
Kirill
Dave Hansen wrote:
On Wed, 2008-04-23 at 10:40 +0400, Kirill Korotaev wrote:
Having all this functionality in a signle
If the current interface is insufficient, we should first expand it in
such a way that it can be used for checkpoint. That certainly won't
work in all cases. fork(), for instance, doesn't take any arguments and
is going to be awfully hard to expand. :)
I'd love to hear some of your
Can you please cat /proc/meminfo after loading OVZ kernel w/o NVIDIA driver?
It's strange and OVZ shouldn't allocate so much memory...
My only guess is that you have a lot of iptables rules which are allocated in
vmalloc.
If not, plz create bug in bugzilla.
Alexander GQ Gerasiov wrote:
Hello
The was no real rationale except for some people seeing clone functionality
as the match and the fact that FS_NAMESCAPE was done so made them believe it is
a good way to go.
And I warned about flags limitation at the beginning.
Both OpenVZ/vserver suggested to use a special syscall for handling
sorry, NFS server is not supported from inside container.
edbch wrote:
Hello everyone
I usually use the forum to solve my problems with OpenVZ, but now I
really need to solve this problem urgently.
I must enable nfs server from one of virtual hosts. To be more precise
must activates a
Cedric Le Goater wrote:
Hello Kirill !
Kirill Korotaev wrote:
Pierre,
my point is that after you've added interface set IPCID, you'll need
more and more for checkpointing:
- create/setup conntrack (otherwise connections get dropped),
- set task start time (needed for Oracle
more clear this time.
Thanks,
Kirill
Pierre Peiffer wrote:
Kirill Korotaev wrote:
Why user space can need this API? for checkpointing only?
I would say at least for checkpointing... ;) May be someone else may find an
interest about this for something else.
In fact, I'm sure that you have
Ouch... Looks like vps_dumpable flag should be moved to task_struct
to fix it properly...
Your fix is incorrect for a number of reasons:
1. it introduces a race: in do_exit() task first looses it's mm, only then
it's exit_state is set to EXIT_ZOMBIE, so there is still a window when it
will
AFAIR it can be ignored.
it warn's that some kernel thread with virtual pid (i.e. from inside VE)
changes it's session/pgrp.
BTW, In 2.6.18 (what kernel are you running?) these lines are changed to:
WARN_ON(__is_virtual_pid(pgrp));
WARN_ON(__is_virtual_pid(session));
Thanks,
Jakob Goldbach wrote:
On Wed, 2008-01-09 at 11:55 +0300, Kirill Korotaev wrote:
AFAIR it can be ignored.
Great.
BTW, In 2.6.18 (what kernel are you running?) these lines are changed to:
WARN_ON(__is_virtual_pid(pgrp));
WARN_ON(__is_virtual_pid(session));
I am
Eric Keller wrote:
Is it possible to start a kernel thread and then move it to a particular
VE?
I have the following code inside of a kernel thread:
envid_t _veid = 200;
// enter that VE
unsigned flags = VE_ENTER;
int err = real_env_create(_veid, flags, 0, 0, 0); // the last 3
From: Cedric Le Goater [EMAIL PROTECTED]
Sorry for the delay, I somehow managed to overlook your message.
This patch is the first step in making the refrigerator() available
to all architectures, even for those without power management.
The purpose of such a change is to be able to use
Daniel Lezcano wrote:
Denis V. Lunev wrote:
Recently David Miller and Herbert Xu pointed out that struct net becomes
overbloated and un-maintainable. There are two solutions:
- provide a pointer to a network subsystem definition from struct net.
This costs an additional dereferrence
- place
imho panic() is too much.
create_singlethread_workqueue() can fail e.g. due to out of memory...
Thanks,
Kirill
Daniel Lezcano wrote:
Subject: make netns cleanup to run in a separate queue
From: Benjamin Thery [EMAIL PROTECTED]
This patch adds a separate workqueue for cleaning up a network
Ah, sorry. Didn't notice it's called only on boot.
Acked-By: Kirill Korotaev [EMAIL PROTECTED]
Kirill Korotaev wrote:
imho panic() is too much.
create_singlethread_workqueue() can fail e.g. due to out of memory...
Thanks,
Kirill
Daniel Lezcano wrote:
Subject: make netns cleanup
imho very good idea and we'll have more and more docs there...
(the only issue I see - it would be nice to have cgroups docs in the same place,
though cgroups are not about namespaces directly.)
Acked-By: Kirill Korotaev [EMAIL PROTECTED]
Pavel Emelyanov wrote:
Hi guys!
As you might have
I dislike this patch:
it's not scalable/efficient to travers all the tasks
while we know the pid namespace we care about.
Kirill
Eric W. Biederman wrote:
This patch implements task_in_pid_ns and uses it to limit cap_set_all
and sys_kill(-1,) to only those tasks in the current pid namespace.
99% of the checkpoining code is concentrated in kernel/cpt/*
Kirill
Yi Wang wrote:
Hello,
I'm new to OpenVZ. I'm interested in learning more about the way OpenVZ
handles its VPS migration, especially the virtualized network stack.
I just downloaded the linux-2.6.18-openvz source code.
yes, OpenVZ calculates idle statistics for VE.
it is calculated as follows:
cpu user/system time are incremented as usual on timer tick,
idle time is incremented on per-VE per-VCPU runqueue when there is no tasks
in VE runqueue.
i.e. VE idle time is the time when there were no VE tasks on the
Virtualization of sysv msg queues is incomplete:
msg_hdrs and msg_bytes variables visible from userspace are global.
Let's make them per-namespace.
Signed-Off-By: Alexey Kuznetsov [EMAIL PROTECTED]
Signed-Off-By: Kirill Korotaev [EMAIL PROTECTED]
---
include/linux/ipc.h |2 ++
ipc/msg.c
Eric W. Biederman wrote:
Greg KH [EMAIL PROTECTED] writes:
Also fun is that the dev file implementation needs to be able to
report different major:minor numbers based on which mount of
sysfs we are dealing with.
Um, no, that's not going to happen. /dev/sda will _always_ have the
same
main()
{
unsigned long c;
c = read_cr3_a();
write_cr3(c | 0x80);
c = read_cr3_b();
write_cr3(c | 0x100);
}
-- cut
Kirill
Arjan van de Ven wrote:
On Tue, 02 Oct 2007 18:08:32 +0400
Kirill Korotaev [EMAIL
Cedric,
how safe does it intersect with netlinks from network namespace?
I see mqueues can send netlink messages, have you checked how safe it is?
Thanks,
Kirill
Cedric Le Goater wrote:
From: Cedric Le Goater [EMAIL PROTECTED]
This patch adds a struct mq_namespace holding the common
marked as volatile already.
I faced this bug myself in i686 arch code when did code
rearrangement in 2.6.18.
Signed-Off-By: Kirill Korotaev [EMAIL PROTECTED]
Acked-By: Pavel Emelianov [EMAIL PROTECTED]
---
asm-i386/system.h |2 +-
asm-x86_64/system.h |8
2 files changed, 5
Cedric Le Goater wrote:
* possible direction for C/R user api
. checkpoint/restart syscalls
. C/R file systems
solves the set id issue
elegant but exposes too much the ABI
I vote for the filesystem :) I'd add more details as we did on
Cedric Le Goater wrote:
Many of these were discussed in a recent Zap paper present in USENIX:
http://www.ncl.cs.columbia.edu/publications/usenix2007_fordist.pdf
The paper describes important design choices in Zap (but I'm biased ...).
I think it may serve as an appetizer for the discussion :P
Acked-By: Kirill Korotaev [EMAIL PROTECTED]
[EMAIL PROTECTED] wrote:
From: Daniel Lezcano [EMAIL PROTECTED]
Doing this makes loopback.c a better example of how to do a
simple network device, and it removes the special case
single static allocation of a struct net_device, hopefully
making
Serge E. Hallyn wrote:
Quoting Kir Kolyshkin ([EMAIL PROTECTED]):
Serge E. Hallyn wrote:
I just sent the next (final) version of the containers development
roadmap. Eric Biederman and Paul Menage will be at kernel summit
representing this work.
We had planned on having a containers
Andi Kleen wrote:
Not everyone likes frame buffer
You don't need the frame buffer; cards typically have text mode
fonts upto 80x50. The node numbers vary, but you can find out yours
with vga=ask
but even with it any OOPs in
network code which happens in softirq, io scheduler and
Daniel Pittman wrote:
[EMAIL PROTECTED] writes:
Should we include this in the patchset ?
[...]
Only the global-init process must be special - any other
container-init process must be killable to prevent run-away processes
in the system.
One problem I hit while using OpenVZ is
Oleg Nesterov wrote:
On 08/01, Dave Hansen wrote:
If the main thread is exiting, but is not the last thread in the
group, should we let it exit and let the next thread in the group
the reaper of the pid ns ?
Well, what happens with a multithreaded init today?
As it was
Serge E. Hallyn wrote:
Quoting Pavel Emelyanov ([EMAIL PROTECTED]):
[snip]
| Maybe it's worth disabling cross-namespaces ptracing...
I think so too. Its probably not a serious limitation ?
Several people think we will implement 'namespace entering' through a
ptrace hack, where maybe the
Oleg Nesterov wrote:
On 07/26, Pavel Emelyanov wrote:
Make task release its namespaces after it has reparented all his
children to child_reaper, but before it notifies its parent about
its death.
The reason to release namespaces after reparenting is that when task
exits it may send a signal to
Unfortunately I'm unlikely to be there (late notice - I really need to
keep a better list of upcoming conferences and proposal deadlines!).
But it still sounds like a good idea.
It's a pity :/ I though you are the primary candidate for presenting containers
at KLS.
Depending on the format
Paul Menage wrote:
On 7/10/07, Serge E. Hallyn [EMAIL PROTECTED] wrote:
A (still under construction) list of features we expect to be worked on
next year looks like this:
4. task containers functionality
specific containers
A couple of more container subsystem
Paul Menage wrote:
On 7/2/07, Serge E. Hallyn [EMAIL PROTECTED] wrote:
4. task containers functionality
How about if we adopt process containers or task containers as the
term for the generic container framework, to distinguish from more
general user-space containers? In the same
Paul Menage wrote:
On 7/12/07, Kirill Korotaev [EMAIL PROTECTED] wrote:
Not sure why it requires some additional controller, but surely
it is possible to create a match for iptables matching container ID.
But which container ID? Don't forget that a task is in one container
in each
Serge E. Hallyn wrote:
(If you missed earlier parts of this thread, you can catch earlier parts of
this thread starting at
https://lists.linux-foundation.org/pipermail/containers/2007-July/005860.html)
Thanks for all the recent feedback. I particularly added a lot from Paul
Menage and
Jan Kara wrote:
Hello,
On Mon 02-07-07 18:16:09, Kirill Korotaev wrote:
it looks like the following fix:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=43c3e6f5abdf6acac9b90c86bf03f995bf7d3d92
was lost after resurrecting of the spliced checkpointing list
Paul Jackson wrote:
Serge wrote:
So if any development is planned over the next year on top of
Pauls Menage's containers and/or cpusets, we'd like that listed
here as well.
Well, apparenntly you guys are planning some container stuff
that is on top of or relates to in some way unclear to
Jan,
it looks like the following fix:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=43c3e6f5abdf6acac9b90c86bf03f995bf7d3d92
was lost after resurrecting of the spliced checkpointing list in this patch:
Ben Greear wrote:
Kirill Korotaev wrote:
Patrick McHardy wrote:
I believe OpenVZ stores the current namespace somewhere global,
which avoids passing the namespace around. Couldn't you do this
as well?
yes, we store a global namespace context on current
(can be stored in per-cpu
Eric W. Biederman wrote:
Patrick McHardy [EMAIL PROTECTED] writes:
Eric W. Biederman wrote:
-- The basic design
There will be a network namespace structure that holds the global
variables for a network namespace, making those global variables
per network namespace.
One of those per network
Patrick McHardy wrote:
Eric W. Biederman wrote:
-- The basic design
There will be a network namespace structure that holds the global
variables for a network namespace, making those global variables
per network namespace.
One of those per network namespace global variables will be the
Ben Greear wrote:
Patrick McHardy wrote:
Eric W. Biederman wrote:
-- The basic design
There will be a network namespace structure that holds the global
variables for a network namespace, making those global variables
per network namespace.
One of those per network namespace global
Jeff Garzik wrote:
Eric W. Biederman wrote:
Jeff Garzik [EMAIL PROTECTED] writes:
David Miller wrote:
I don't accept that we have to add another function argument
to a bunch of core routines just to support this crap,
especially since you give no way to turn it off and get
that function
filesystem_type has have FS_VIRTUALIZED flag to allow
file system to be mountable from inside VE.
However, make sure that code in spufs doesn't have global
variables or at least uses them in a safe manner.
(e.g. it is safe to have global lock for all VEs,
but may be unsafe to have global list of
Paul Menage wrote:
On 6/22/07, Balbir Singh [EMAIL PROTECTED] wrote:
The problem with input in bytes is that the user will have to ensure
that the input is
a multiple of page size, which implies that she would need to use the
calculator every time.
Having input in bytes seems pretty
Balbir Singh wrote:
Kirill Korotaev wrote:
Paul Menage wrote:
On 6/22/07, Balbir Singh [EMAIL PROTECTED] wrote:
The problem with input in bytes is that the user will have to ensure
that the input is
a multiple of page size, which implies that she would need to use the
calculator every time
Christian,
there is still a lot to do in mainstream.
currently mainstream includes ipc and utsname namespaces, pid namespace
is being actively worked on, networking is worked on as well.
Also -mm tree includes containers patches which allow to group tasks
and which will be used for resource
Ingo Molnar wrote:
i'd still like to hear back from Kirill co whether this framework is
flexible enough for their work (OpenVZ, etc.) too.
My IMHO is that so far the proposed group scheduler doesn't look ready/suitable.
We need to have a working SMP version before it will be clear
whether
Deniel,
Daniel Lezcano wrote:
Pavel Emelianov wrote:
I did this at the very first version, but Alexey showed me that this
would be wrong. Look. When we create the second device it must be in
the other namespace as it is useless to have them in one namespace.
But if we have the device in the
My apologies, but I won't be able to visit OLS this year :/
I think Kirill Kolyshkin and Denis Lunev will be glad to meet you again!
Hope Pavel Emelianov will be able to join as well.
Thanks,
Kirill
Serge E. Hallyn wrote:
Last year we all met for breakfast at OLS. Now we've all pretty much
http://git.openvz.org/?p=linux-2.6.18-openvz;a=commit;h=cb649b7cede6764c00e256578dc3c7ad73c1b24c
Thanks,
Kirill
Christian Kaiser2 wrote:
Hi Kirill,
imho there is no reason for not adding the patch to the git repository.
I've tested it for one week now and I'm getting no serious errors.
Cedric,
just a small note.
imho it is not correct to check performance with enabled debug in memory
allocator
since it can influence cache efficiency much.
In you case looks like you have DEBUG_SLAB enabled.
Pavel will recheck as well what influences on this particular test.
BTW, it is
the results were also very reproducible but the profiling was too noisy.
we also changed the kernel. the previous pidns patchset was on a 2.6.21-mm2
and we ported it on a 2.6.22-rc1-mm1.
If reproducible, then were they the same as Pavel posted?
but let me remove some debugging options,
Christian,
sys_fairsched_chwt() can return ENOENT in the only case:
when fairsched node ID was not found in the list.
but VE0 node is always on the list.
So I guess there is something wrong with syscall args in your case.
Please apply the attached patch to the kernel and vzctl
recompile/reboot
Christian Kaiser2 wrote:
Kirill Korotaev [EMAIL PROTECTED] wrote on 05/25/2007 02:41:09 PM:
Christian,
sys_fairsched_chwt() can return ENOENT in the only case:
when fairsched node ID was not found in the list.
but VE0 node is always on the list.
So I guess there is something wrong
Christian, please check the patch attached.
Kirill
Christian Kaiser2 wrote:
Kirill Korotaev [EMAIL PROTECTED] wrote on 05/25/2007 02:41:09 PM:
Christian,
sys_fairsched_chwt() can return ENOENT in the only case:
when fairsched node ID was not found in the list.
but VE0 node is always
Srivatsa Vaddagiri wrote:
On Fri, May 25, 2007 at 05:05:16PM +0400, Kirill Korotaev wrote:
That way the scheduler would first pick a virtual CPU to schedule, and
then pick a user from that virtual CPU, and then a task from the user.
don't you mean the vice versa:
first use to scheduler
looks like you have 32bit binaries in vzctl since configure script didn't add
-m64 to CFLAGS, though it has a code in configure.ac for this.
Checking...
Kirill
Christian Kaiser2 wrote:
Hi all,
I did my first steps wtih OpenVZ on a JS21 Blade. I'm running FC6 and built
my own 2.6.18 kernel
Andrew Morton wrote:
On Thu, 17 May 2007 23:20:12 +0530
Balbir Singh [EMAIL PROTECTED] wrote:
A meaningful container size does not hamper performance. I am in the process
of getting more results (with varying container sizes). Please let me know
what you think of the results? Would you like
Balbir Singh wrote:
This patch is inspired by the discussion at http://lkml.org/lkml/2007/4/11/187
and implements per container statistics as suggested by Andrew Morton
in http://lkml.org/lkml/2007/4/11/263. The patch is on top of 2.6.21-mm1
with Paul's containers v9 patches (forward ported)
Cedric Le Goater wrote:
Hello !
Whenever I reply to an email sent on the openvz mailing the To: used
is devel@openvz.org, which is ok, but the person who it was from, is
not in
the Cc: ...
This feels wrong. Can you confirm the behavior and eventually fix it ?
Cedric,
In fact the mailman
Kirill,
We are working on getting a few UTS namespace tests included in Linux
Testsuite Project (LTP) and were wondering if you (or anyone else) had
any test cases for IPC namespace.
If there are tests for IPC NS, can you share those tests with us ?
And, do you have any plans to push
Dave Hansen wrote:
On Wed, 2007-04-11 at 11:59 +0400, Kirill Korotaev wrote:
Dave, taskref sounds a bit too much generic for me...
I completely agree. It's a pretty generic name. In the kernel, though
it does provide lookups to tasks. I think the in-kernel task vs.
process naming
The loss of performances is very noticeable inside the container and
seems to be directly related to the usage of the pair device and the
specific network configuration needed for the container. When the
packets are sent by the container, the mac address is for the pair
device but the IP address
. Biederman wrote:
Kirill Korotaev [EMAIL PROTECTED] writes:
we have the hack below in ip_forward() to avoid skb_cow(),
Banjamin, can you check whether it helps in your case please?
(NOTE: you will need to replace check for NETIF_F_VENET with something else
or introduce the same flag on etun
Eric W. Biederman wrote:
Benjamin Thery [EMAIL PROTECTED] writes:
My investigations on the increase of cpu load when running netperf inside a
container (ie. through etun2-etun1) is progressing slowly.
I'm not sure the cause is fragmentation as we supposed initially.
In fact, it seems related
Eric W. Biederman wrote:
Daniel Lezcano [EMAIL PROTECTED] writes:
Benjamin Thery and I we were looking at this.
For the moment we are investigating if there is IP fragmentation between the
eth0 and the pair devices.
The profiling shows us pskb_expand_head and csum_partial_copy_generic
unwind library
that top level frame is different of syscall frame. It is the case
unless process was born with CLONE_STOPPED.
Author: Alexey Kuznetsov [EMAIL PROTECTED]
Signed-Off-By: Alexey Kuznetsov [EMAIL PROTECTED]
Signed-Off-By: Kirill Korotaev [EMAIL PROTECTED]
--- a/arch/ia64/kernel/unwind.c
1. You can always reboot to the previous kernel (until you replaced
or removed it). Keep in mind for future, that keeping previous stable
kernel is always a good idea.
2. I guess you probably:
a) have incorrect .config file.
b) or you somehow incorrectly compiled/installed the kernel.
How
Nick,
Accounting becomes easy if we have a container pointer in struct page.
This can form base ground for building controllers since any memory
related controller would be interested in tracking pages. However we
still want to evaluate if we can build them without bloating the
struct page.
So what to do when virtual physical limit is hit?
OOM-kill current task?
when the RSS limit is hit, but there _are_ enough
pages left on the physical system, there is no
good reason to swap out the page at all
- there is no benefit in doing so (performance
wise, that is)
- it
Eric,
And misses every resource sharing opportunity in sight.
that was my point too.
Except for
filtering the which pages are eligible for reclaim an RSS limit should
not need to change the existing reclaim logic, and with things like the
memory zones we have had that kind of restriction in
Herbert,
Just curious why current vserver code kills arbitrary
task in container then?
because it obviously lacks the finess of OpenVZ code :)
seriously, handling the OOM kills inside a container
has never been a real world issue, as once you are
really out of memory (and OOM starts
Eric W. Biederman wrote:
Pavel Emelianov [EMAIL PROTECTED] writes:
Adds needed pointers to mm_struct and page struct,
places hooks to core code for mm_struct initialization
and hooks in container_init_early() to preinitialize
RSS accounting subsystem.
An extra pointer in struct page is
On Mon, 2007-03-12 at 19:23 +0300, Kirill Korotaev wrote:
For these you essentially need per-container page-_mapcount counter,
otherwise you can't detect whether rss group still has the page in question
being mapped
in its processes' address spaces or not.
What do you mean by this? You
Andrew Morton wrote:
On Tue, 06 Mar 2007 17:55:29 +0300
Pavel Emelianov [EMAIL PROTECTED] wrote:
+struct rss_container {
+ struct res_counter res;
+ struct list_head page_list;
+ struct container_subsys_state css;
+};
+
+struct page_container {
+ struct page *page;
+
Here was a thread related to this:
http://forum.openvz.org/index.php?t=treeth=479mid=2771rev=reveal=
Thanks,
Kirill
Shuveb Hussain wrote:
Hi,
Is there a way to obtain the current CPU utilization for a particular VPS?
How do I figure out CPU utilization:
1. utilization with reference to
Pavel Emelianov wrote:
Balbir Singh wrote:
Pavel Emelianov wrote:
This patchset adds RSS, accounting and control and
limiting the number of tasks and files within container.
Based on top of Paul Menage's container subsystem v7
RSS controller includes per-container RSS accounter,
reclamation
Acked-By: Kirill Korotaev [EMAIL PROTECTED]
From: Serge E. Hallyn [EMAIL PROTECTED]
Subject: [PATCH] namespaces: update some function names
The {get,exit}_task_namespaces do not grab references to the individual
namespaces, only to the nsproxy. Reflect that in the function names.
Not so
87 matches
Mail list logo