headers
Kir Kolyshkin | 3 Aug 2008 21:42
Favicon

Re: resource limits don't apply immediately on 2.6.24

Jan Kokoska wrote:
> Hi,
>
> I'm used to having resource limits apply immediately on 2.6.18 kernel
> releases of OpenVZ. I run "vzctl set VEID --privvmpages soft:hard
> --save" and "free" in the virtual machine reports the new figure right
> away.
>
> This is not the case on 2.6.24-19-openvz Ubuntu kernel, where
> user_beancounters report the newer figure and the virtual machine
> would report the newer figure if I restarted it, however the change
> doesn't apply online.
>
> Is this a feature or a bug?
>   
First, the thing you need to check is not what free shows -- look at the 
appropriate line in /proc/user_beancounters.

What free shows is some kind of approximation, and if it doesn't work 
right away first check that vzctl version you use in both cases is the 
same. Then check for 'meminfo' setting in VE config. Then report a bug 
to bugzilla.openvz.org
Permalink | Reply | 0 comments |
headers
Kir Kolyshkin | 3 Aug 2008 21:46
Favicon

Re: SOLVED: Re: How to kill dead container (init is dead)

Pongracz Istvan wrote:
> Hi,
>
> Andrey helped me a lot about this issue and it seems, it was a human
> error:
> I cancelled a backup process and this container left in suspended state.
>
> I used the vzdump script to make backups and I cancelled it while it
> started to create an rsync.
>   

I think that vzdump can be guarded with traps to restore container state 
in case backup is interrupted. In bash there's a 'trap' command which 
should help.

> So, be careful :)
>
> Cheers, 
> IStván
>
> 2008. 07. 30, szerda keltezéssel 22.16-kor Pongracz Istvan ezt írta:
>   
>> Hi,
>>
>> I use openvz kernel 2.6.18-028stab051 for long months on my gentoo
>> system.
>> The uptime now is 105 days.
>>
>> It seems, one of my containers completely dead: 
>> all processes are dead, including the init process.
(Continue reading)

Permalink | Reply | 2 comments |
headers
Pongracz Istvan | 3 Aug 2008 23:53
Picon

Re: Re: How to kill dead container (init is dead)

2008. 08. 3, vasárnap keltezéssel 23.46-kor Kir Kolyshkin ezt írta:
> Pongracz Istvan wrote:
> > Hi,
> >
> > Andrey helped me a lot about this issue and it seems, it was a human
> > error:
> > I cancelled a backup process and this container left in suspended state.
> >
> > I used the vzdump script to make backups and I cancelled it while it
> > started to create an rsync.
> >   
> 
> I think that vzdump can be guarded with traps to restore container state 
> in case backup is interrupted. In bash there's a 'trap' command which 
> should help.
> 

Hi,

Thank you for your help!
I found something about the vixie-cron + ssmtp incompatibility on the
net.
When I start the script manually, it is working well everytime.
But when it started from cron, it failed, almost in 100%.

Some tests are still running...... I will report again :)

Thank you!

István
(Continue reading)

Permalink | Reply | 1 comment |
headers
Garith Dugmore | 5 Aug 2008 17:10
Picon

Re: simfs 2tb limit?

Hi Benoit and Arjan,

Thanks for your replies - I've been away from my Inbox.

[Benoit Branciard]
> If possible without critical data loss or downtime, repartitioning 
> would be an option to consider. You may for example allocate some 
> space (say 70GB) on sda for the container's roots (mounted on /vz), 
> and keep the huge /data to provide data space to containers by means 
> of bind mounts on various points of their diretory tree.
[Garith]

Ideally I'd like to not use partitioning as this limits the flexibility 
I have with changing the amount of disk space a container has access to 
but it might be the solution in this scenario.

[Arjan]

I must say I haven't tried with quotas, but without quotas OpenVZ can 
handle it perfectly.
My container has DISK_QUOTA="no" set, and it can use my 5 TB perfectly:

# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/vzfs             4.8T  2.2T  2.7T  45% /
#

So I guess you should look into the limits of the quota support.

[Garith]
(Continue reading)

Permalink | Reply | 2 comments |
headers
Adeel Nazir | 6 Aug 2008 08:16
Picon
Favicon

Re: simfs 2tb limit?


--- On Tue, 8/5/08, Garith Dugmore <garith@...> wrote:

> From: Garith Dugmore <garith@...>
> Subject: Re: [Users] simfs 2tb limit?
> To: users@...
> Date: Tuesday, August 5, 2008, 11:10 AM
> Hi Benoit and Arjan,
> 
> Thanks for your replies - I've been away from my Inbox.
> 
> [Benoit Branciard]
> > If possible without critical data loss or downtime,
> repartitioning 
> > would be an option to consider. You may for example
> allocate some 
> > space (say 70GB) on sda for the container's roots
> (mounted on /vz), 
> > and keep the huge /data to provide data space to
> containers by means 
> > of bind mounts on various points of their diretory
> tree.
> [Garith]
> 
> Ideally I'd like to not use partitioning as this limits
> the flexibility 
> I have with changing the amount of disk space a container
> has access to 
> but it might be the solution in this scenario.
>
(Continue reading)

Permalink | Reply | 1 comment |
headers
Pongracz Istvan | 8 Aug 2008 08:15
Picon

[Really SOLVED] Re: How to kill dead container (init is dead)

2008. 08. 3, vasárnap keltezéssel 23.53-kor Pongracz Istvan ezt írta:
> > 
> 
> Hi,
> 
> Thank you for your help!
> I found something about the vixie-cron + ssmtp incompatibility on the
> net.

Bingo.

> When I start the script manually, it is working well everytime.
> But when it started from cron, it failed, almost in 100%.
> 
> Some tests are still running...... I will report again :)

Hi All,

After several days I can confirm, the problem is gone.
vzdump succeed every night.

<background for the search engines>
The basic problem was, vixie-cron and ssmtp have problems if they are
working together.
I run several maintenance tasks regularly on the hardware node, such as
daily backup with vzdump, logwatch, check rootkits etc.
I want to see email reports about that, so, I need a 3rd party rescue
smtp server. I choosed gmail with ssmtp.
I expected, vzdump cannot complete its task and I got stopped container
etc.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Garith Dugmore | 8 Aug 2008 11:45
Picon

Re: simfs 2tb limit?

Hi Adeel,

Thanks for this - will keep it in mind.

Cheers!

Adeel Nazir wrote:
You could consider using LVM on the raid array, to give yourself dynamic partitioning, so you can increase the 'partition' when needed. Bear in mind that LVM adds a little overhead, but I think the flexibility you get with LVM makes up for the overhead.
--
Garith Dugmore Senior Systems Administrator South African Astronomical Observatory and Southern African Large Telescope Phone: 021 460 9343 Fax: 021 447 3639 SAAO Website: http://www.saao.ac.za SALT Website: http://www.salt.ac.za Jabber: gar_I_th-2ASvDZBniIcQZ7m9OUuf1w@public.gmane.org MSN: garithd-PkbjNfxxIARBDgjK7y7TUQ@public.gmane.org Skype: garith5512 
_______________________________________________
Users mailing list
Users@...
https://openvz.org/mailman/listinfo/users
Permalink | Reply | 0 comments |
headers
Marcin Owsiany | 9 Aug 2008 04:25
Picon
Gravatar

/etc/hosts mangling on restarts

Hi,

If you have already seen that, then I apologize for the additional
noise, but in case this was overlooked, I would like to point out my new
comment to bug #855, which has all the details about $SUBJECT:

http://bugzilla.openvz.org/show_bug.cgi?id=855#c4

I think my points make it somewhat more severe than the original report.
Either way, I would love to hear what developers think about it.

regards,

--

-- 
Marcin Owsiany <marcin@...>              http://marcin.owsiany.pl/
GnuPG: 1024D/60F41216  FE67 DA2D 0ACA FC5E 3F75  D6F6 3A0D 8AA0 60F4 1216

"Every program in development at MIT expands until it can read mail."
                                                              -- Unknown
Permalink | Reply | 0 comments |
headers
Aleksandr Levchuk | 12 Aug 2008 23:10
Picon
Gravatar

Re: No source code in download.openvz.org/debian/

Ola, Kir,

Thank you very much!

Alex

On Mon, Jul 28, 2008 at 4:20 AM, Ola Lundqvist <ola@...> wrote:
> Hi Aleksandr
>
> The source exist in Debian as the packages:
> linux-patch-openvz (or kernel-patch-openvz)
> linux-source-2.6.24 (or linux-source-2.6.18)
>
> See the description on how to compile is in the linux-patch-openvz package
> as the README.Debian.
>
> The configuration file is included in the deb package that you can download.
>
> Best regards,
>
> // Ola
>
> On Mon, Jul 28, 2008 at 12:30:54PM +0400, Kir Kolyshkin wrote:
>> Ola,
>>
>> Can you please help Aleksandr?
>>
>> Aleksandr Levchuk wrote:
>> >Dear download.openvz.org Maintainer,
>> >
>> >For this Debain package
>> >http://download.openvz.org/debian/dists/etch/main/binary-amd
>> >64/base/linux-image-2.6.24-openvz-24-004.1d1-amd64_004.1d1_a md64.deb
>> >would it be possible to publish the source?
>> >
>> >Only Kernel of 2.6.18 and higher support the 34 Infiniband cards
>> >(mlx4_core) that I have and
>> >I need it to compile a module that is not included in the above DEB
>> >package (igb2xxx).
>> >
>> >
>> >The expected location to find the sources is
>> >http://download.openvz.org/debian/dists/etch/main/source/, but there
>> >is nothing there.
>> >
>> >Alex
>> >
>> >
>> >
>>
>>
>
> --
>  --- Inguza Technology AB --- MSc in Information Technology ----
> /  ola@...                    Annebergsslingan 37        \
> |  opal@...                   654 65 KARLSTAD            |
> |  http://inguza.com/                Mobile: +46 (0)70-332 1551 |
> \  gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9  /
>  ---------------------------------------------------------------
>

--

-- 
--------------------------------------------
Aleksandr Levchuk
University of California, Riverside
1-951-368-0004
--------------------------------------------
Permalink | Reply | 0 comments |
headers
Kir Kolyshkin | 13 Aug 2008 02:47
Favicon

Re: [Debian] vzlist

Papp Tamás wrote:
> Hi all,
>
> How can be moved a VE to an other machine?
>
> The files are on drbd.

I think you need to do the following:
1. vzctl chkpnt 100
2. copy the dump file (it's in /vz/dump/), if it's not under drbd.
3. copy the config file
4. vzctl restore 100 (on a destination machine)

PS it makes more sense to use users <at>  list for such questions

> I umounted it, removed the openvz kernel modules, but vzlist still 
> shows the VE.
>
> Also I mount the drbd on another system and copied 100.conf to 
> /etc/vz/conf/ but didn't help.
>
> What does vzlist and vzctl tools does look?
>
> The system is Ubunti 7.10 with own kernel:  2.6.24-ovz002-drbd8.0.11 .
Permalink | Reply | 0 comments |

Gmane