Xen users - Aug 2009 - increase size for dom guest using lvm online

could we increase the size for dom guest online without rebooting or unmounting?

pls kindly advise any methods.



      Yahoo!香港提供網上安全攻略，教你如何防範黑客! 請前往 http://hk.promo.yahoo.com/security/ 了解更多!

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Aclhk Aclhk wrote:
>could we increase the size for dom guest online without rebooting or 
>unmounting?
I don''t believe so. At the moment, I believe the size passed to the 
guest is fixed at creating time, so if you enlarge the volume, the 
guest won''t get to know about it until it is shut down and 
re-created. It may be possible to detatch a volume and re-attach it 
with xm but I''ve never done that.

In any case, at least with ext3, you need the filesystem unmounted in 
order to resize it.

-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> 
> could we increase the size for dom guest online without rebooting or
> unmounting?
> 
> pls kindly advise any methods.
> 
For a PV DomU (or a HVM DomU using PV drivers), there is no mechanism
for the backend (Dom0) to notify the frontend (DomU) of a change in
device size. I suspect that HVM DomU''s have the same problems.

So no, it cannot be done without rebooting.

James


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Just for the sake of it: ext3 has supported online expansion for some time 
now.

-Bart
----- Original Message ----- 
From: "Simon Hobson" <linux@thehobsons.co.uk>
To: "Aclhk Aclhk" <aclhkaclhk@ymail.com>;
<xen-users@lists.xensource.com>
Sent: Sunday, August 16, 2009 1:23 PM
Subject: Re: [Xen-users] increase size for dom guest using lvm online

> Aclhk Aclhk wrote:
>>could we increase the size for dom guest online without rebooting or 
>>unmounting?
>
> I don''t believe so. At the moment, I believe the size passed to
the guest
> is fixed at creating time, so if you enlarge the volume, the guest
won''t
> get to know about it until it is shut down and re-created. It may be 
> possible to detatch a volume and re-attach it with xm but I''ve
never done
> that.
>
> In any case, at least with ext3, you need the filesystem unmounted in 
> order to resize it.
>
> -- 
> Simon Hobson
>
> Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
> author Gladys Hobson. Novels - poetry - short stories - ideal as
> Christmas stocking fillers. Some available as e-books.
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
> 

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

"James Harper" <james.harper@bendigoit.com.au> writes:
>> could we increase the size for dom guest online without rebooting or
>> unmounting?
>> 
>> pls kindly advise any methods.
>
> For a PV DomU (or a HVM DomU using PV drivers), there is no mechanism
> for the backend (Dom0) to notify the frontend (DomU) of a change in
> device size. I suspect that HVM DomU''s have the same problems.
>
> So no, it cannot be done without rebooting.
Unless you resort to double attach and replacing device-mapper
tables.  Possible in principle, but be very careful suspending your
root device...  Good luck!
-- 
Feri.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> At the moment, I believe the size passed to the guest is fixed at 
> creating time, so if you enlarge the volume, the guest won''t get
to
> know about it until it is shut down and re-created.
I just verified that this is the case on Xen 3.2 on Debian 5.  I was not 
able to get the guest to pick up the change in device size.

There is the option of doing LVM in the guest, though this will make the 
volumes more difficult to access from the Xen dom0.  If you do lvm in 
the guest you will be able to attach a new chunk of disk, then run 
pvcreate on it and add it to a volume group in your guest.  However if 
the dom0 is running lvm as well it will try to register the volume 
groups defined by the PVs that you are intending to map through to your 
guests.  You have two options for preventing this from creating problems 
on the dom0.  First, you could exclude from the scan process in lvm.conf 
those devices destined to be used with LVM in a guest, but you will not 
be able to mount these volumes in the dom0 as a result.  The other 
option is to name your volume groups differently for every guest so 
there are no conflicts when lvm in the dom0 does a pvscan, this allows 
you to use the volumes from dom0 but can be a manageability issue.

Personally I don''t think the tradeoffs are worth it for online resize.
I do my LVM on the dom0 and then send individual LVs to the guests as 
xvda, xvdb, etc. but I cannot resize these without a reboot.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Sun, Aug 16, 2009 at 6:25 PM, James
Harper<james.harper@bendigoit.com.au> wrote:
>>
>> could we increase the size for dom guest online without rebooting or
>> unmounting?
>
> For a PV DomU (or a HVM DomU using PV drivers), there is no mechanism
> for the backend (Dom0) to notify the frontend (DomU) of a change in
> device size. I suspect that HVM DomU''s have the same problems.
>
> So no, it cannot be done without rebooting.

There''s a workaround, actually. For PV domU anyway.
While you cannot get domU to pickup device size change online, you CAN
attach new block device. So if you use LVM on domU as well, you can
simply attach a new block device, add it as new PV, then extend domU''s
LV.

So you get the ability to grow domU''s storage online at the cost of
extra complexity (having LVM on domU).

-- 
Fajar

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

I wrote:
>It may be possible to detatch a volume and re-attach it with xm but 
>I''ve never done that.
Well now I have. Advance planning says I''ll need a bit more space on 
a volume, and there was a little bit of space available on the iSCSI 
box (so a case of grab it while it''s there, before someone else does 
!).

Unmounted the volume in the guest, then :

# cat backup
<stuff deleted>
disk    = [ ''phy:vg0/backuproot,xvda1,w'', 
''phy:vg0/backupvar,xvda2,w'', 
''phy:/dev/disk/by-path/ip-<very-long-iscsi-address>,xvda3,w'',
''phy:vg0/backupswap,xvda4,w'' ]

# xm list
Name                                        ID   Mem VCPUs      State   Time(s)
Domain-0                                     0   486     4     r-----  67982.4
backup                                      11   511     2     -b---- 135441.2

# xm block-detach 11 xvda3

<do any other steps required, such as altering partitions or resizing 
LVM volumes>

# xm block-attach 11
''/dev/disk/by-path/ip-<very-long-iscsi-address>''
/dev/xvda3 r
Error: Device 51715 (vbd) could not be connected. Hotplug scripts not working.
Usage: xm block-attach <Domain> <BackDev> <FrontDev>
<Mode> [BackDomain]

Create a new virtual block device.

Checked in teh guest and the device didn''t re-appear, so tried again :

# xm block-attach 11
''/dev/disk/by-path/ip-<very-long-iscsi-address>''
/dev/xvda3 r
Error: Device /dev/xvda3 (51715, vbd) is already connected.
Usage: xm block-attach <Domain> <BackDev> <FrontDev>
<Mode> [BackDomain]

Create a new virtual block device.


So I still had to shutdown and re-create the guest for the size 
change to appear.


>In any case, at least with ext3, you need the filesystem unmounted 
>in order to resize it.
As "Bart Van Kerckhove" <bart@it-ss.be> pointed out, you can do 
online resize of an ext3 filesystem :

# df -h /store
Filesystem            Size  Used Avail Use% Mounted on
/dev/xvda3             99G   44G   51G  47% /store

# resize2fs /dev/xvda3
resize2fs 1.41.3 (12-Oct-2008)
Filesystem at /dev/xvda3 is mounted on /store; on-line resizing required
old desc_blocks = 7, new_desc_blocks = 10
Performing an on-line resize of /dev/xvda3 to 39321087 (4k) blocks.
The filesystem on /dev/xvda3 is now 39321087 blocks long.

# df -h /store
Filesystem            Size  Used Avail Use% Mounted on
/dev/xvda3            148G   44G   97G  31% /store
-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Wed, Aug 19, 2009 at 04:17:41PM +0100, Simon Hobson
wrote:
> I wrote:
> 
> >It may be possible to detatch a volume and re-attach it with xm but 
> >I''ve never done that.
> 
> Well now I have. Advance planning says I''ll need a bit more space
on
> a volume, and there was a little bit of space available on the iSCSI 
> box (so a case of grab it while it''s there, before someone else
does
> !).
> 
> Unmounted the volume in the guest, then :
> 
> # cat backup
> <stuff deleted>
> disk    = [ ''phy:vg0/backuproot,xvda1,w'', 
> ''phy:vg0/backupvar,xvda2,w'', 
>
''phy:/dev/disk/by-path/ip-<very-long-iscsi-address>,xvda3,w'',
> ''phy:vg0/backupswap,xvda4,w'' ]
> 
> # xm list
> Name                                        ID   Mem VCPUs      State   
> Time(s)
> Domain-0                                     0   486     4     r-----  
> 67982.4
> backup                                      11   511     2     -b---- 
> 135441.2
> 
> # xm block-detach 11 xvda3
> 
> <do any other steps required, such as altering partitions or resizing 
> LVM volumes>
> 
> # xm block-attach 11
''/dev/disk/by-path/ip-<very-long-iscsi-address>''
> /dev/xvda3 r
> Error: Device 51715 (vbd) could not be connected. Hotplug scripts not 
> working.
> Usage: xm block-attach <Domain> <BackDev> <FrontDev>
<Mode> [BackDomain]
> 
> Create a new virtual block device.
> 
> Checked in teh guest and the device didn''t re-appear, so tried
again :
> 
> # xm block-attach 11
''/dev/disk/by-path/ip-<very-long-iscsi-address>''
> /dev/xvda3 r
> Error: Device /dev/xvda3 (51715, vbd) is already connected.
> Usage: xm block-attach <Domain> <BackDev> <FrontDev>
<Mode> [BackDomain]
> 
> Create a new virtual block device.
> 
> 
> So I still had to shutdown and re-create the guest for the size 
> change to appear.
>
You need to attach a completely new device, not the same xvda3.

-- Pasi
 
> 
> 
> >In any case, at least with ext3, you need the filesystem unmounted 
> >in order to resize it.
> 
> As "Bart Van Kerckhove" <bart@it-ss.be> pointed out, you
can do
> online resize of an ext3 filesystem :
> 
> # df -h /store
> Filesystem            Size  Used Avail Use% Mounted on
> /dev/xvda3             99G   44G   51G  47% /store
> 
> # resize2fs /dev/xvda3
> resize2fs 1.41.3 (12-Oct-2008)
> Filesystem at /dev/xvda3 is mounted on /store; on-line resizing required
> old desc_blocks = 7, new_desc_blocks = 10
> Performing an on-line resize of /dev/xvda3 to 39321087 (4k) blocks.
> The filesystem on /dev/xvda3 is now 39321087 blocks long.
> 
> # df -h /store
> Filesystem            Size  Used Avail Use% Mounted on
> /dev/xvda3            148G   44G   97G  31% /store
> -- 
> Simon Hobson
> 
> Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
> author Gladys Hobson. Novels - poetry - short stories - ideal as
> Christmas stocking fillers. Some available as e-books.
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Pasi Kärkkäinen wrote:
>  > # xm block-attach 11
''/dev/disk/by-path/ip-<very-long-iscsi-address>''
>>  /dev/xvda3 r
>>  Error: Device 51715 (vbd) could not be connected. Hotplug scripts not
>>  working.
>>  Usage: xm block-attach <Domain> <BackDev> <FrontDev>
<Mode> [BackDomain]
>>
>>  Create a new virtual block device.
>>
>>  Checked in teh guest and the device didn''t re-appear, so
tried again :
>>
>>  # xm block-attach 11
''/dev/disk/by-path/ip-<very-long-iscsi-address>''
>>  /dev/xvda3 r
>>  Error: Device /dev/xvda3 (51715, vbd) is already connected.
>>  Usage: xm block-attach <Domain> <BackDev> <FrontDev>
<Mode> [BackDomain]
>>
>>  Create a new virtual block device.
>>
>>
>>  So I still had to shutdown and re-create the guest for the size
>>  change to appear.
>>
>
>You need to attach a completely new device, not the same xvda3.
Are you saying it''s not possible to use the same 
device name that was previously detatched ?

-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users