Xen users - Apr 2008 - Convert CentOS system to Xen image?

I have a CentOS server that I started to rebuild using Xen. It''s
turning
out to have lot more configuring than I thought. So instead of building
a new server and setting everything up again, is there any way to
convert a running, stand alone linux system into a xen image?

Googling I see people converting xen to parallels, vmware to xen, and
even Windows to Xen, but I could not find converting a linux system. 

Thanks,
James


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

On Di, Apr 01, 2008 at 02:49:43 -0400, James Pifer
wrote:
> I have a CentOS server that I started to rebuild using Xen. It''s
turning
> out to have lot more configuring than I thought. So instead of building
> a new server and setting everything up again, is there any way to
> convert a running, stand alone linux system into a xen image?
> 

Yes, of course you can.



First of all, you must know, do you want to convert it
to a PV or an HVM Xen domain.

HVM:

1) Just copy image of your server to a logical volume of 
the Xen server LVM or to any other block device.

Xen uses plain raw images for vm images storing.

To do this you can boot your Cent OS to single user mode (and bring
network up manually) or use Knoppix or any live cd.

2) Create config for new domain.

3) Run it.



PV:

Like HVM, but after that (and before domain start),
you also should do:

1) Copy into domain file system kernel modules 
   (/lib/modules/`uname -r`)

2) Modify /etc/fstab (if needed)

3) Modify /etc/udev/rules.d (if needed)




It''s better to start from converting into HVM
and the convert it into PV if needed.

> Googling I see people converting xen to parallels, vmware to xen, and
> even Windows to Xen, but I could not find converting a linux system. 
> 
> Thanks,
> James
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
-- 
WBR, i.m.chubin


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

> First of all, you must know, do you want to convert it
> to a PV or an HVM Xen domain.
I believe I need to convert to a PV. 
> 
> HVM:
> 
> 1) Just copy image of your server to a logical volume of 
> the Xen server LVM or to any other block device.
Is it possible to create an uncompressed image using something like g4l
and make use of that image file somehow?

Thanks,
James



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

James Pifer wrote:
> > First of all, you must know, do you want to convert it
> > to a PV or an HVM Xen domain.
> 
> I believe I need to convert to a PV. 
> 
> > 
> > HVM:
> > 
> > 1) Just copy image of your server to a logical volume of 
> > the Xen server LVM or to any other block device.
> 
> Is it possible to create an uncompressed image using something like g4l
> and make use of that image file somehow?
Watch out for device name changes, like sda->xvda, in you
configuration.

How big is the volumes of the source system?

If they are really big it might be best to kickstart a fresh
install into a PV, then import the configuration and packages
from the source system to the new PV.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

On Di, Apr 01, 2008 at 03:47:20 -0400, James Pifer
wrote:
> > First of all, you must know, do you want to convert it
> > to a PV or an HVM Xen domain.
> 
> I believe I need to convert to a PV. 
> 
> > 
> > HVM:
> > 
> > 1) Just copy image of your server to a logical volume of 
> > the Xen server LVM or to any other block device.
> 
> Is it possible to create an uncompressed image using something like g4l
> and make use of that image file somehow?
I think yes, but I have never used it.


You can create image with cat, netcat or ssh.


knoopix#  cat /dev/sda | nc -l -p 3000

dom0#     nc knoppix 3000 > centos.img   

or 

(if you use LVM and have a volume group called VG)

dom0#     lvcreate -L 20G -n centos /dev/VG
dom0#     nc knoppix 3000 > /dev/VG/centos

> 
> Thanks,
> James
> 
> 
-- 
WBR, i.m.chubin


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

On Tue, 2008-04-01 at 15:54 -0400, Ross S. W. Walker
wrote:
> James Pifer wrote:
> > > First of all, you must know, do you want to convert it
> > > to a PV or an HVM Xen domain.
> > 
> > I believe I need to convert to a PV. 
> > 
> > > 
> > > HVM:
> > > 
> > > 1) Just copy image of your server to a logical volume of 
> > > the Xen server LVM or to any other block device.
> > 
> > Is it possible to create an uncompressed image using something like
g4l
> > and make use of that image file somehow?
> 
> Watch out for device name changes, like sda->xvda, in you
> configuration.
> 
> How big is the volumes of the source system?
> 
> If they are really big it might be best to kickstart a fresh
> install into a PV, then import the configuration and packages
> from the source system to the new PV.
> 
Not that bit at all. Total used space on the system is about 8 gig, and
I might be able to trim that down. But I don''t have free space to
create
a separate logical drive. So I would need to do this in an image file. 

Thanks,
James


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

> I think yes, but I have never used it.
> 
> 
> You can create image with cat, netcat or ssh.
> 
> 
> knoopix#  cat /dev/sda | nc -l -p 3000
> 
> dom0#     nc knoppix 3000 > centos.img   
> 
Any chance of this working while the system is up, instead of booting
with knoppix (or anything else)? Or would too much not be copied because
it''s locked?

One of my other challenges is that this system is remote to me so I
would have to ask (beg) someone to give me hand. Also, being a live
production system, downtime is never easy...

Never used netcat before, that looks pretty cool. 

Thanks,
James


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

On Di, Apr 01, 2008 at 04:17:19 -0400, James Pifer
wrote:
> > I think yes, but I have never used it.
> > 
> > 
> > You can create image with cat, netcat or ssh.
> > 
> > 
> > knoopix#  cat /dev/sda | nc -l -p 3000
> > 
> > dom0#     nc knoppix 3000 > centos.img   
> > 
> 
> Any chance of this working while the system is up, instead of booting
> with knoppix (or anything else)? Or would too much not be copied because
> it''s locked?

The reason not in locks actually, 
but better to say in the fact that copied filesystem is being
modified while copying. Hence it''s likely that it will be
inconsistent after copying finished (I mean target fs).


If you can remount all of the source file systems to ro mode,
you may not reboot to live cd.

Also, if it is impossible, 
you can shut all processes (or better to switch to init 1, 
but as far as I understand you want to ssh to the server, so 
sshd must be running as minimum), that can modify filesystem, 
and after that make

# sync

and start copying.


Also you can use ssh instead of nc (ssh eats more CPU power due to
crypting, but it is not so important in your case).

centos# sync ; cat /dev/hda | ssh dom0 '' cat > centos.img ''



> 
> One of my other challenges is that this system is remote to me so I
> would have to ask (beg) someone to give me hand. Also, being a live
> production system, downtime is never easy...
> 
> Never used netcat before, that looks pretty cool. 
> 
> Thanks,
> James
> 
-- 
WBR, i.m.chubin


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

Attempted to do this using netcat + ssh and here''s my next problem. The
CentOS server is using LVM, so instead of getting an img file of around
8 gig that I expected, I got one that is 18 gig, or the size of the LV. 

I tried it anyway and I get a valid boot sector not found. 

Any suggestions?

Thanks,
James 


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

James Pifer wrote:
> Attempted to do this using netcat + ssh and here''s my next
problem. The
> CentOS server is using LVM, so instead of getting an img file of around
> 8 gig that I expected, I got one that is 18 gig, or the size of the LV. 
> 
> I tried it anyway and I get a valid boot sector not found. 
> 
> Any suggestions?
Don''t do netcat, create a fresh PV with the same installed packages
as on the old machine and copy the configuration of the old server
that is pertinent. You could look at creating a kickstart file to
have those packages installed during the initial install, or just
install base/core and add in by hand the missing packages via yum.

For data partitions like /home you can do a dump, sftp, and restore.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

On Wed, Apr 2, 2008 at 8:41 AM, James Pifer <jep@obrien-pifer.com> wrote:
> Attempted to do this using netcat + ssh and here''s my next
problem. The
> CentOS server is using LVM, so instead of getting an img file of around
> 8 gig that I expected, I got one that is 18 gig, or the size of the LV.
>
> I tried it anyway and I get a valid boot sector not found.
>
> Any suggestions?

Xen uses a partition within a partiton.  If you are going from a real system
to a xen and they are both on lvms do something like

lvcreate -n xenBoot -L <a little bigger than the real servers lvm>G
/dev/VolGroup00
fdisk /dev/VolGroup00/xenBoot
  create a primary partion

kpartx -a /dev/VolGroup/xendBoot  <this will make the partition visible to
dom0 in /dev/mapper/xenBoot1

dd if=/dev/VolGroup/<your real boot volume> of=/dev/mapper/xenBoot1


>
>
> Thanks,
> James
>
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>


-- 
Joe Royall
407-401-2007


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

On Mi, Apr 02, 2008 at 08:41:40 -0400, James Pifer
wrote:
> Attempted to do this using netcat + ssh and here''s my next
problem. The
> CentOS server is using LVM, so instead of getting an img file of around
> 8 gig that I expected, I got one that is 18 gig, or the size of the LV. 

If you use netcat (and dd), you will get image
with the size of underlying harddisk.

(not LV, but whole disk).


What command have you use to copy image?

If you use line

    cat /dev/DEVICE | nc ...

what do you specify as the DEVICE?

> 
> I tried it anyway and I get a valid boot sector not found. 
> 
> Any suggestions?
> 
> Thanks,
> James 
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
-- 
WBR, i.m.chubin


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

On Mi, Apr 02, 2008 at 09:19:19 -0400, Ross S. W. Walker
wrote:
> James Pifer wrote:
> > Attempted to do this using netcat + ssh and here''s my next
problem. The
> > CentOS server is using LVM, so instead of getting an img file of
around
> > 8 gig that I expected, I got one that is 18 gig, or the size of the
LV.
> > 
> > I tried it anyway and I get a valid boot sector not found. 
> > 
> > Any suggestions?
> 
> Don''t do netcat, create a fresh PV with the same installed
packages
> as on the old machine and copy the configuration of the old server
> that is pertinent. You could look at creating a kickstart file to
> have those packages installed during the initial install, or just
> install base/core and add in by hand the missing packages via yum.

You can use dd or cat (+ netcat | ssh) 
and it will copy raw image,

but creating LVM and filesystems on its logical volumes
has several advantages:

1. You can filesystems with any layout and sizes
2. You transfer only data and not "freespace" as in case of dd/netcat
usage


Disatvantage:

1. You should create LVM and FS first. 

> 
> For data partitions like /home you can do a dump, sftp, and restore.
Also you can use rsync
(it should be installed on both hosts).


> 
> -Ross
> 
> ______________________________________________________________________
> This e-mail, and any attachments thereto, is intended only for use by
> the addressee(s) named herein and may contain legally privileged
> and/or confidential information. If you are not the intended recipient
> of this e-mail, you are hereby notified that any dissemination,
> distribution or copying of this e-mail, and any attachments thereto,
> is strictly prohibited. If you have received this e-mail in error,
> please immediately notify the sender and permanently delete the
> original and any copy or printout thereof.
> 
-- 
WBR, i.m.chubin


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

Igor Chubin wrote:
> On Mi, Apr 02, 2008 at 09:19:19 -0400, Ross S. W. Walker wrote:
> > James Pifer wrote:
> > > Attempted to do this using netcat + ssh and here''s my
next problem. The
> > > CentOS server is using LVM, so instead of getting an img file of
around
> > > 8 gig that I expected, I got one that is 18 gig, or the size of
the LV.
> > >
> > > I tried it anyway and I get a valid boot sector not found.
> > >
> > > Any suggestions?
> >
> > Don''t do netcat, create a fresh PV with the same installed
packages
> > as on the old machine and copy the configuration of the old server
> > that is pertinent. You could look at creating a kickstart file to
> > have those packages installed during the initial install, or just
> > install base/core and add in by hand the missing packages via yum.
>
>
> You can use dd or cat (+ netcat | ssh)
> and it will copy raw image,
Yes, but that makes a lot of assumptions about the source and
destination systems that might not be true, like, the source
system is physical and the destination virtual, so the
destination will have different disk names, the source may
have software RAID1, but the destination being virtual may
already be backed by RAID1 and not need it. The source disk
may be infinitely bigger then the allocated destination
(ie source drive 250GB, root partition is 16GB LV and only
3GB of it is used, but netcat''ing will try to transfer full
250GB into a 16 or 32GB guest volume).
> but creating LVM and filesystems on its logical volumes
> has several advantages:
>
> 1. You can filesystems with any layout and sizes
> 2. You transfer only data and not "freespace" as in case of
> dd/netcat usage
>
>
> Disatvantage:
>
> 1. You should create LVM and FS first.
>
I don''t think I grasp your advantages/disadvantages points
here, maybe it''s a language barrier where the subtleties
were lost in translation.

I do agree that LVM is the way to go, as it''s pros far
outweigh it''s cons.
> >
> > For data partitions like /home you can do a dump, sftp, and restore.
>
> Also you can use rsync
> (it should be installed on both hosts).
rsync will indiscriminately copy-over files that it shouldn''t.

No this is why companies sell stand-alone P2V products.

You can look for these, but if you had just created the PV and
duplicated the important configuration it''d be up and running
and in testing phase now.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

On Wed, 2008-04-02 at 16:32 +0300, Igor Chubin wrote:
> On Mi, Apr 02, 2008 at 08:41:40 -0400, James Pifer wrote:
> > Attempted to do this using netcat + ssh and here''s my next
problem. The
> > CentOS server is using LVM, so instead of getting an img file of
around
> > 8 gig that I expected, I got one that is 18 gig, or the size of the
LV.
> 
> 
> If you use netcat (and dd), you will get image
> with the size of underlying harddisk.
> 
> (not LV, but whole disk).
> 
> 
> What command have you use to copy image?
> 
> If you use line
> 
>     cat /dev/DEVICE | nc ...
> 

I used:
cat /dev/ida/c0d0p2 | nc -l -p 3000

Then on dom0 I used:
netcat centoshost 3000 > centos.img

James


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

...
> > You can use dd or cat (+ netcat | ssh)
> > and it will copy raw image,
> 
> Yes, but that makes a lot of assumptions about the source and
> destination systems that might not be true, like, the source
> system is physical and the destination virtual, so the
> destination will have different disk names, the source may
> have software RAID1, but the destination being virtual may
> already be backed by RAID1 and not need it. The source disk
> may be infinitely bigger then the allocated destination
> (ie source drive 250GB, root partition is 16GB LV and only
> 3GB of it is used, but netcat''ing will try to transfer full
> 250GB into a 16 or 32GB guest volume).

I agree with you.
My advice is related to particular (described earlier) task.

...
> > Disatvantage:
> >
> > 1. You should create LVM and FS first.
> >
> 
> I don''t think I grasp your advantages/disadvantages points
> here, maybe it''s a language barrier where the subtleties
> were lost in translation.
My mother tongue is Russian and my English is very very poor.
It''s really hard to quickly express my thoughts in English.
Excuse me Ross, and all who read this (and other my postings to the list).

When I say that dd/cat + netcat usage has advantage over 
LVM+FS+rsync/scp/dump usage in the particular case, 
I mean:

when you use cat/dd you can copy system quick and dirty, 
but when you use high level tools for copying files, 
you should create volumes and filesystems before copying.

I stress, that cat/dd one should use only
in particular cases (for "quick-and-dirty" migration). 
And for full featured migration
it''s better to use LVM/FS and other high-level tools.
> I do agree that LVM is the way to go, as it''s pros far
> outweigh it''s cons.
>
Ross, I think that I fully agree with you
(except the fact that I think that there are many
tasks where dd/cat is enough).
> > >
> > > For data partitions like /home you can do a dump, sftp, and
restore.
> >
> > Also you can use rsync
> > (it should be installed on both hosts).
> 
> rsync will indiscriminately copy-over files that it shouldn''t.
What do you mean?

-- 
WBR, i.m.chubin


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

On Mi, Apr 02, 2008 at 10:03:04 -0400, James Pifer
wrote:
> On Wed, 2008-04-02 at 16:32 +0300, Igor Chubin wrote:
> > On Mi, Apr 02, 2008 at 08:41:40 -0400, James Pifer wrote:
> > > Attempted to do this using netcat + ssh and here''s my
next problem. The
> > > CentOS server is using LVM, so instead of getting an img file of
around
> > > 8 gig that I expected, I got one that is 18 gig, or the size of
the LV.
> > 
> > 
> > If you use netcat (and dd), you will get image
> > with the size of underlying harddisk.
> > 
> > (not LV, but whole disk).
> > 
> > 
> > What command have you use to copy image?
> > 
> > If you use line
> > 
> >     cat /dev/DEVICE | nc ...
> > 
> 
> 
> I used:
> cat /dev/ida/c0d0p2 | nc -l -p 3000
You have copied just one partition,
instead of the whole disk.

To copy the whole disk,
you should use:

# cat /dev/ida/c0d0 | nc -l -p 3000

But this command will copy all of the disk (and if it is big,
it is not good idea; also you need enough freespace on the target
system).

More flexible way is the way mentioned by Ross: 
to use LVM on the target system
and to copy not the raw device, but files, which is situated
on the device.



> 
> Then on dom0 I used:
> netcat centoshost 3000 > centos.img
> 
> James
> 
-- 
WBR, i.m.chubin


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

Igor Chubin wrote:
> ...
> > > You can use dd or cat (+ netcat | ssh)
> > > and it will copy raw image,
> > 
> > Yes, but that makes a lot of assumptions about the source and
> > destination systems that might not be true, like, the source
> > system is physical and the destination virtual, so the
> > destination will have different disk names, the source may
> > have software RAID1, but the destination being virtual may
> > already be backed by RAID1 and not need it. The source disk
> > may be infinitely bigger then the allocated destination
> > (ie source drive 250GB, root partition is 16GB LV and only
> > 3GB of it is used, but netcat''ing will try to transfer full
> > 250GB into a 16 or 32GB guest volume).
> 
> I agree with you.
> My advice is related to particular (described earlier) task.
> 
> ...
> > > Disatvantage:
> > >
> > > 1. You should create LVM and FS first.
> > >
> > 
> > I don''t think I grasp your advantages/disadvantages points
> > here, maybe it''s a language barrier where the subtleties
> > were lost in translation.
> 
> My mother tongue is Russian and my English is very very poor.
> It''s really hard to quickly express my thoughts in English.
> Excuse me Ross, and all who read this (and other my postings 
> to the list).
> 
> When I say that dd/cat + netcat usage has advantage over 
> LVM+FS+rsync/scp/dump usage in the particular case, 
> I mean:
> 
> when you use cat/dd you can copy system quick and dirty, 
> but when you use high level tools for copying files, 
> you should create volumes and filesystems before copying.
> 
> I stress, that cat/dd one should use only
> in particular cases (for "quick-and-dirty" migration). 
> And for full featured migration
> it''s better to use LVM/FS and other high-level tools.
> 
> > I do agree that LVM is the way to go, as it''s pros far
> > outweigh it''s cons.
> >
> 
> Ross, I think that I fully agree with you
> (except the fact that I think that there are many
> tasks where dd/cat is enough).
> 
> > > >
> > > > For data partitions like /home you can do a dump, sftp, and
restore.
> > >
> > > Also you can use rsync
> > > (it should be installed on both hosts).
> > 
> > rsync will indiscriminately copy-over files that it
shouldn''t.
> 
> What do you mean?
Well rsync was really designed to replicate data from
one system to another, but this isn''t technically
replication here as not all configs will be the same.

If it''s a data volume like /home, then you might get
by with rsync. I have found though that the time to
get rsync going efficiently doesn''t pay for a one
time operation. If this is done over a secure
reliable network and not the Internet, you could do
a dump | ssh restore type deal, maybe add a gzip
step in there to compress over the wire and it''s
done.

I have been playing around with the idea of using
something like ''revisor'' from Fedora to capture
the basic system settings and packages into a
kickstart install iso that you can use to install
a PV that''s identical to the source. Have it do
an RPM audit of all config files except a few
key ones like fstab, ifcfg etc and such and tar
those up and have it extracted in the %post
section of the kickstart... It''s an interesting
idea... Hmmm

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

I simply commented out the line of boot in /etc/fstab.

hope it works.

On Wed, Apr 2, 2008 at 8:41 AM, James Pifer <jep@obrien-pifer.com> wrote:
> Attempted to do this using netcat + ssh and here''s my next
problem. The
> CentOS server is using LVM, so instead of getting an img file of around
> 8 gig that I expected, I got one that is 18 gig, or the size of the LV.
>
> I tried it anyway and I get a valid boot sector not found.
>
> Any suggestions?
>
> Thanks,
> James
>
>
> _______________________________________________
> 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

...
> > > > > For data partitions like /home you can do a dump, sftp,
and restore.
> > > >
> > > > Also you can use rsync
> > > > (it should be installed on both hosts).
> > > 
> > > rsync will indiscriminately copy-over files that it
shouldn''t.
> > 
> > What do you mean?
> 
> Well rsync was really designed to replicate data from
> one system to another, but this isn''t technically
> replication here as not all configs will be the same.
> 
> If it''s a data volume like /home, then you might get
> by with rsync. I have found though that the time to
> get rsync going efficiently doesn''t pay for a one
> time operation. If this is done over a secure
> reliable network and not the Internet, you could do
> a dump | ssh restore type deal, maybe add a gzip
> step in there to compress over the wire and it''s
> done.

If I understand you right, 
you say that


hostA# dump / | ssh hostB restore

is more effective than:

hostA# rsync -a / hostB:/

Or you''ve said 
«
> by with rsync. I have found though that the time to
> get rsync going efficiently doesn''t pay for a one
»
about setup time?

rsync installation is very quick.
It''s enough to simply install rsync package
and then you can run it (it uses ssh as transport).

> I have been playing around with the idea of using
> something like ''revisor'' from Fedora to capture
> the basic system settings and packages into a
> kickstart install iso that you can use to install
> a PV that''s identical to the source. Have it do
> an RPM audit of all config files except a few
> key ones like fstab, ifcfg etc and such and tar
> those up and have it extracted in the %post
> section of the kickstart... It''s an interesting
> idea... Hmmm


Really interesting idea. Thank you for it.


I''m looking for the way how I can 
automagically (without rsyncing, dumping and another big
data transfers) create clone of a system 
from standard repositories.


Your idea is to package configs delta between
current system (that we want to clone) and fresh system
installed from repo state.


Cloned system configs - fresh system configs => delta.rpm


And after installing delta.rpm into fresh system 
we will have system that will be equal to cloned.

Correct?


> 
> -Ross
> 
> ______________________________________________________________________
> This e-mail, and any attachments thereto, is intended only for use by
> the addressee(s) named herein and may contain legally privileged
> and/or confidential information. If you are not the intended recipient
> of this e-mail, you are hereby notified that any dissemination,
> distribution or copying of this e-mail, and any attachments thereto,
> is strictly prohibited. If you have received this e-mail in error,
> please immediately notify the sender and permanently delete the
> original and any copy or printout thereof.
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
-- 
WBR, i.m.chubin


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

On Wed, Apr 2, 2008 at 9:52 AM, Igor Chubin <igor@chub.in>
wrote:
>  If I understand you right,
>  you say that
>
>
>  hostA# dump / | ssh hostB restore
>
>  is more effective than:
>
>  hostA# rsync -a / hostB:/
i like to use rsync to replicate running servers.  of course, for
one-shot copy it''s no faster than other dump-transfer-restore schemas;
but it''s quite easy to use to quickly converge almost to a
''snapshot''
capability:

1: do a full rsync, might take tooo long, lots of stuff wouldn''t be
really consistent
2: repeat rsync, should be far faster, copying only the differences.
3: keep repeating until the whole copy is comfortably short. usually
the third iteration is less than a couple minutes
4: quick: shut down services, one last rsync, start services.
5: done!


-- 
Javier

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

Igor Chubin wrote:
> ...
> > > > > > For data partitions like /home you can do a dump,
sftp, and restore.
> > > > >
> > > > > Also you can use rsync
> > > > > (it should be installed on both hosts).
> > > > 
> > > > rsync will indiscriminately copy-over files that it
shouldn''t.
> > > 
> > > What do you mean?
> > 
> > Well rsync was really designed to replicate data from
> > one system to another, but this isn''t technically
> > replication here as not all configs will be the same.
> > 
> > If it''s a data volume like /home, then you might get
> > by with rsync. I have found though that the time to
> > get rsync going efficiently doesn''t pay for a one
> > time operation. If this is done over a secure
> > reliable network and not the Internet, you could do
> > a dump | ssh restore type deal, maybe add a gzip
> > step in there to compress over the wire and it''s
> > done.
> 
> 
> If I understand you right, 
> you say that
> 
> 
> hostA# dump / | ssh hostB restore
> 
> is more effective than:
> 
> hostA# rsync -a / hostB:/
> 
> Or you''ve said 
> <
> > by with rsync. I have found though that the time to
> > get rsync going efficiently doesn''t pay for a one
> >
> about setup time?
> 
> rsync installation is very quick.
> It''s enough to simply install rsync package
> and then you can run it (it uses ssh as transport).
I was unaware that you didn''t need to setup an rsync
daemon on the other end and configure security.

Then sure I guess rsync would be quicker, does it
preserve special files, pipes, symlinks and sparse
files properly?
> > I have been playing around with the idea of using
> > something like ''revisor'' from Fedora to capture
> > the basic system settings and packages into a
> > kickstart install iso that you can use to install
> > a PV that''s identical to the source. Have it do
> > an RPM audit of all config files except a few
> > key ones like fstab, ifcfg etc and such and tar
> > those up and have it extracted in the %post
> > section of the kickstart... It''s an interesting
> > idea... Hmmm
> 
> Really interesting idea. Thank you for it.
> 
> 
> I''m looking for the way how I can 
> automagically (without rsyncing, dumping and another big
> data transfers) create clone of a system 
> from standard repositories.
> 
> 
> Your idea is to package configs delta between
> current system (that we want to clone) and fresh system
> installed from repo state.
> 
> 
> Cloned system configs - fresh system configs => delta.rpm
> 
> 
> And after installing delta.rpm into fresh system 
> we will have system that will be equal to cloned.
> 
> Correct?
Well my idea for the configs was more rudimentary, but
your idea of a delta RPM is perfect! What is the tool
used for creating delta RPMs? All one would need is
to create a list of configs that should never be
packaged and create a delta RPM of all the others...

Then once you have the delta RPM you add it to the
list of RPMs included in the revisor spin and voila!

Hmmm, I suppose when it comes to the actual data you could
do a number of things like tar or cpio to a series of
DVD images. Maybe some dvd authoring tool can help automate
this process... You would need some interesting method to
find all the data though as not all data is in easily
identifiable locations like home. An example of this
would be the samba tdb files like passdb.tdb and
secrets.tdb, which are created after install and aren''t
included in the catalog.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

Javier Guerra wrote:
> On Wed, Apr 2, 2008 at 9:52 AM, Igor Chubin <igor@chub.in> wrote:
> >  If I understand you right,
> >  you say that
> >
> >
> >  hostA# dump / | ssh hostB restore
> >
> >  is more effective than:
> >
> >  hostA# rsync -a / hostB:/
> 
> i like to use rsync to replicate running servers.  of course, for
> one-shot copy it''s no faster than other dump-transfer-restore
schemas;
> but it''s quite easy to use to quickly converge almost to a
''snapshot''
> capability:
> 
> 1: do a full rsync, might take tooo long, lots of stuff wouldn''t
be
> really consistent
> 2: repeat rsync, should be far faster, copying only the differences.
> 3: keep repeating until the whole copy is comfortably short. usually
> the third iteration is less than a couple minutes
> 4: quick: shut down services, one last rsync, start services.
> 5: done!
Or, take a snapshot, mount it, then rsync off that.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

...
> > Or you''ve said 
> > <
> > > by with rsync. I have found though that the time to
> > > get rsync going efficiently doesn''t pay for a one
> > >
> > about setup time?
> > 
> > rsync installation is very quick.
> > It''s enough to simply install rsync package
> > and then you can run it (it uses ssh as transport).
> 
> I was unaware that you didn''t need to setup an rsync
> daemon on the other end and configure security.

You need daemon in anonymous mode only.
Except this mode, rsync uses transport (ssh by default
and -e /some/path/to/program if specified).

> 
> Then sure I guess rsync would be quicker, does it
> preserve special files, pipes, symlinks and sparse
> files properly?

Yes.
All of them (use -a key for that).

> 
> > > I have been playing around with the idea of using
> > > something like ''revisor'' from Fedora to capture
> > > the basic system settings and packages into a
> > > kickstart install iso that you can use to install
> > > a PV that''s identical to the source. Have it do
> > > an RPM audit of all config files except a few
> > > key ones like fstab, ifcfg etc and such and tar
> > > those up and have it extracted in the %post
> > > section of the kickstart... It''s an interesting
> > > idea... Hmmm
> > 
> > Really interesting idea. Thank you for it.
> > 
> > 
> > I''m looking for the way how I can 
> > automagically (without rsyncing, dumping and another big
> > data transfers) create clone of a system 
> > from standard repositories.
> > 
> > 
> > Your idea is to package configs delta between
> > current system (that we want to clone) and fresh system
> > installed from repo state.
> > 
> > 
> > Cloned system configs - fresh system configs => delta.rpm
> > 
> > 
> > And after installing delta.rpm into fresh system 
> > we will have system that will be equal to cloned.
> > 
> > Correct?
> 
> Well my idea for the configs was more rudimentary, but
> your idea of a delta RPM is perfect! What is the tool
> used for creating delta RPMs? All one would need is
> to create a list of configs that should never be
> packaged and create a delta RPM of all the others...
I''m not rpm guru (I use debian), but I have just googled
and found what we need (at least it''s very like to it).


The tool name is deltarpm :)

More info here [1].



[1]
http://www-uxsup.csx.cam.ac.uk/pub/doc/suse/suse9.3/suselinux-adminguide_en/sec.rpm.html
> 
> Then once you have the delta RPM you add it to the
> list of RPMs included in the revisor spin and voila!

Yes.
> 
> Hmmm, I suppose when it comes to the actual data you could
> do a number of things like tar or cpio to a series of
> DVD images. Maybe some dvd authoring tool can help automate
> this process... You would need some interesting method to
> find all the data though as not all data is in easily
> identifiable locations like home. An example of this
> would be the samba tdb files like passdb.tdb and
> secrets.tdb, which are created after install and aren''t
> included in the catalog.


The idea was about configs, not about data actually.


Of course, if you want to migrate file server or DBMS system, 
rpm delta is not enough.

> 
> -Ross
> 
> ______________________________________________________________________
> This e-mail, and any attachments thereto, is intended only for use by
> the addressee(s) named herein and may contain legally privileged
> and/or confidential information. If you are not the intended recipient
> of this e-mail, you are hereby notified that any dissemination,
> distribution or copying of this e-mail, and any attachments thereto,
> is strictly prohibited. If you have received this e-mail in error,
> please immediately notify the sender and permanently delete the
> original and any copy or printout thereof.
> 
-- 
WBR, i.m.chubin


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

On Mi, Apr 02, 2008 at 11:18:42 -0400, Ross S. W. Walker
wrote:
> Javier Guerra wrote:
> > On Wed, Apr 2, 2008 at 9:52 AM, Igor Chubin <igor@chub.in>
wrote:
> > >  If I understand you right,
> > >  you say that
> > >
> > >
> > >  hostA# dump / | ssh hostB restore
> > >
> > >  is more effective than:
> > >
> > >  hostA# rsync -a / hostB:/
> > 
> > i like to use rsync to replicate running servers.  of course, for
> > one-shot copy it''s no faster than other dump-transfer-restore
schemas;
> > but it''s quite easy to use to quickly converge almost to a
''snapshot''
> > capability:
> > 
> > 1: do a full rsync, might take tooo long, lots of stuff
wouldn''t be
> > really consistent
To solve the problem (if you use LVMs)
you should make snapshots and than sync files 
from them.
> > 2: repeat rsync, should be far faster, copying only the differences.
> > 3: keep repeating until the whole copy is comfortably short. usually
> > the third iteration is less than a couple minutes
> > 4: quick: shut down services, one last rsync, start services.
> > 5: done!
> 
> Or, take a snapshot, mount it, then rsync off that.
Don''t forget to make sync before snapshotting.
It will work without syncing to but it''s better
to use it to be sure that there no information
cached by FS and not written to LVs.

# sync; lvcreate -n snap -L2G /dev/VG/source



> 
> -Ross
> 
> ______________________________________________________________________
> This e-mail, and any attachments thereto, is intended only for use by
> the addressee(s) named herein and may contain legally privileged
> and/or confidential information. If you are not the intended recipient
> of this e-mail, you are hereby notified that any dissemination,
> distribution or copying of this e-mail, and any attachments thereto,
> is strictly prohibited. If you have received this e-mail in error,
> please immediately notify the sender and permanently delete the
> original and any copy or printout thereof.
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
-- 
WBR, i.m.chubin


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

On Wed, Apr 2, 2008 at 10:18 AM, Ross S. W. Walker
<rwalker@medallion.com> wrote:
>
> Javier Guerra wrote:
>  > i like to use rsync to replicate running servers.  of course, for
>  > one-shot copy it''s no faster than other
dump-transfer-restore schemas;
>  > but it''s quite easy to use to quickly converge almost to a
''snapshot''
>  > capability:
>  >
>  > 1: do a full rsync, might take tooo long, lots of stuff
wouldn''t be
>  > really consistent
>  > 2: repeat rsync, should be far faster, copying only the differences.
>  > 3: keep repeating until the whole copy is comfortably short. usually
>  > the third iteration is less than a couple minutes
>  > 4: quick: shut down services, one last rsync, start services.
>  > 5: done!
>
>  Or, take a snapshot, mount it, then rsync off that.
sure, much better... when the original server uses LVM _and_ has free
space on the VG.  otherwise (and quite often) a few rsync iterations
save the day.



-- 
Javier

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

Igor Chubin wrote:
> ...
> > > Or you''ve said 
> > > <
> > > > by with rsync. I have found though that the time to
> > > > get rsync going efficiently doesn''t pay for a one
> > > >
> > > about setup time?
> > > 
> > > rsync installation is very quick.
> > > It''s enough to simply install rsync package
> > > and then you can run it (it uses ssh as transport).
> > 
> > I was unaware that you didn''t need to setup an rsync
> > daemon on the other end and configure security.
> 
> 
> You need daemon in anonymous mode only.
> Except this mode, rsync uses transport (ssh by default
> and -e /some/path/to/program if specified).
> 
> 
> > 
> > Then sure I guess rsync would be quicker, does it
> > preserve special files, pipes, symlinks and sparse
> > files properly?
> 
> 
> Yes.
> All of them (use -a key for that).
Well then I guess rsync is probably the way to go then
in this scenario.

> > > > I have been playing around with the idea of using
> > > > something like ''revisor'' from Fedora to
capture
> > > > the basic system settings and packages into a
> > > > kickstart install iso that you can use to install
> > > > a PV that''s identical to the source. Have it do
> > > > an RPM audit of all config files except a few
> > > > key ones like fstab, ifcfg etc and such and tar
> > > > those up and have it extracted in the %post
> > > > section of the kickstart... It''s an interesting
> > > > idea... Hmmm
> > > 
> > > Really interesting idea. Thank you for it.
> > > 
> > > 
> > > I''m looking for the way how I can 
> > > automagically (without rsyncing, dumping and another big
> > > data transfers) create clone of a system 
> > > from standard repositories.
> > > 
> > > 
> > > Your idea is to package configs delta between
> > > current system (that we want to clone) and fresh system
> > > installed from repo state.
> > > 
> > > 
> > > Cloned system configs - fresh system configs => delta.rpm
> > > 
> > > 
> > > And after installing delta.rpm into fresh system 
> > > we will have system that will be equal to cloned.
> > > 
> > > Correct?
> > 
> > Well my idea for the configs was more rudimentary, but
> > your idea of a delta RPM is perfect! What is the tool
> > used for creating delta RPMs? All one would need is
> > to create a list of configs that should never be
> > packaged and create a delta RPM of all the others...
> 
> I''m not rpm guru (I use debian), but I have just googled
> and found what we need (at least it''s very like to it).
> 
> 
> The tool name is deltarpm :)
> 
> More info here [1].
> 
> 
> 
> [1] 
> http://www-uxsup.csx.cam.ac.uk/pub/doc/suse/suse9.3/suselinux-
> adminguide_en/sec.rpm.html
Thanks, I''ll take a look at that. I don''t know if all
distros support delta RPMs at this point though. :-(

Since your a Debian person, is there any similar tool
to do this on Debian?
> > 
> > Then once you have the delta RPM you add it to the
> > list of RPMs included in the revisor spin and voila!
> 
> 
> Yes.
> 
> > 
> > Hmmm, I suppose when it comes to the actual data you could
> > do a number of things like tar or cpio to a series of
> > DVD images. Maybe some dvd authoring tool can help automate
> > this process... You would need some interesting method to
> > find all the data though as not all data is in easily
> > identifiable locations like home. An example of this
> > would be the samba tdb files like passdb.tdb and
> > secrets.tdb, which are created after install and aren''t
> > included in the catalog.
> 
> 
> 
> The idea was about configs, not about data actually.
True, but some data, like the samba dbs I mentioned would be
essential in preserving the machine''s network identity during
the conversion.

> Of course, if you want to migrate file server or DBMS system, 
> rpm delta is not enough.
No, your right earlier that the actual data migration is too
sensitive to handle here, but we would need to identify what
"extra" configurations should also be bundled in (ie samba),
this can probably be determined by what packages are
installed, or handled en masse.

-Ross


______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

> You have copied just one partition,
> instead of the whole disk.
> 
> To copy the whole disk,
> you should use:
> 
> # cat /dev/ida/c0d0 | nc -l -p 3000
> 
> But this command will copy all of the disk (and if it is big,
> it is not good idea; also you need enough freespace on the target
> system).
> 
> More flexible way is the way mentioned by Ross: 
> to use LVM on the target system
> and to copy not the raw device, but files, which is situated
> on the device.
I do have enough space to do the whole disk, I just don''t have free
space to create an LV. I did it again using /dev/ida/c0d0. Now when I
try to bring up in Xen I get No kernel was found. 

I also downloaded the DVD for the same version of CentOS. So I thought
maybe I would try booting off that and see if I can fix it. I figured if
the iso was mounted as a CDROM first that it would boot, but I still get
No kernel was found.

I''ll try creating a new VM using the DVD, and try to rsync everything,
similar to what Ross was saying. 

The good thing is that I can try lots of things, as long as I don''t
have
to take the other server offline. 

I appreciate everyone''s help!

James


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

On Mi, Apr 02, 2008 at 12:40:00 -0400, James Pifer
wrote:
> > You have copied just one partition,
> > instead of the whole disk.
> > 
> > To copy the whole disk,
> > you should use:
> > 
> > # cat /dev/ida/c0d0 | nc -l -p 3000
> > 
> > But this command will copy all of the disk (and if it is big,
> > it is not good idea; also you need enough freespace on the target
> > system).
> > 
> > More flexible way is the way mentioned by Ross: 
> > to use LVM on the target system
> > and to copy not the raw device, but files, which is situated
> > on the device.
> 
> I do have enough space to do the whole disk, I just don''t have
free
> space to create an LV. I did it again using /dev/ida/c0d0. Now when I
> try to bring up in Xen I get No kernel was found. 
Great!

Now
you should edit kernel command line
and specify new root fs path.

What bootloader to you use?

Presuming you use GRUB:


1. press e in the GRUB bootprompt
2. Choose kernel line
3. Press e again
4. Change root=/dev/ida/c0d0p1 arg to root=/dev/hda1
5. add parameter init=/bin/sh
6. press b to boot
7. when you will get shell prompt do
mount -t proc proc /proc
mount -o rw,remount /
mount -a 
vi /etc/fstab

and here change /dev/ida/c0d0pXXXX to /dev/hdaXXXX

also you can use sed instead of vi:

sed -i s@/dev/ida/c0d0p@/dev/hda@ /etc/fstab

8. umount -a
9. continue booting process
exec /sbin/init




> 
> I also downloaded the DVD for the same version of CentOS. So I thought
> maybe I would try booting off that and see if I can fix it. I figured if
> the iso was mounted as a CDROM first that it would boot, but I still get
> No kernel was found.
I think that when you fix boot loader config and /etc/fstab
you will boot successfully. 
So you are close to the victory.

You even not obliged to use live CD or installation CD to do this.
All can be done be bootloader.


> 
> I''ll try creating a new VM using the DVD, and try to rsync
everything,
> similar to what Ross was saying. 
> 
> The good thing is that I can try lots of things, as long as I
don''t have
> to take the other server offline. 
> 
> I appreciate everyone''s help!
> 
> James
> 
-- 
WBR, i.m.chubin


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

> Great!
> 
> Now
> you should edit kernel command line
> and specify new root fs path.
> 
> What bootloader to you use?
> 
> Presuming you use GRUB:
> 
> 
> 1. press e in the GRUB bootprompt
> 2. Choose kernel line
> 3. Press e again
> 4. Change root=/dev/ida/c0d0p1 arg to root=/dev/hda1
> 5. add parameter init=/bin/sh
> 6. press b to boot
> 7. when you will get shell prompt do
> mount -t proc proc /proc
> mount -o rw,remount /
> mount -a 
> vi /etc/fstab
> 
> and here change /dev/ida/c0d0pXXXX to /dev/hdaXXXX
> 
> also you can use sed instead of vi:
> 
> sed -i s@/dev/ida/c0d0p@/dev/hda@ /etc/fstab
> 
> 8. umount -a
> 9. continue booting process
> exec /sbin/init
> 
Problem is I don''t even get to grub. After I create the new machine in
xen, select the image file for the hard disk, set the other settings, it
gives me a popup that no kernels can be found. Did I miss some steps?

Thanks,
James


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

On Mi, Apr 02, 2008 at 01:35:14 -0400, James Pifer
wrote:
> > Great!
> > 
> > Now
> > you should edit kernel command line
> > and specify new root fs path.
> > 
> > What bootloader to you use?
> > 
> > Presuming you use GRUB:
> > 
> > 
> > 1. press e in the GRUB bootprompt
> > 2. Choose kernel line
> > 3. Press e again
> > 4. Change root=/dev/ida/c0d0p1 arg to root=/dev/hda1
> > 5. add parameter init=/bin/sh
> > 6. press b to boot
> > 7. when you will get shell prompt do
> > mount -t proc proc /proc
> > mount -o rw,remount /
> > mount -a 
> > vi /etc/fstab
> > 
> > and here change /dev/ida/c0d0pXXXX to /dev/hdaXXXX
> > 
> > also you can use sed instead of vi:
> > 
> > sed -i s@/dev/ida/c0d0p@/dev/hda@ /etc/fstab
> > 
> > 8. umount -a
> > 9. continue booting process
> > exec /sbin/init
> > 
> 
> Problem is I don''t even get to grub. After I create the new
machine in
> xen, select the image file for the hard disk, set the other settings, it
> gives me a popup that no kernels can be found. Did I miss some steps?
Please, show the config file of the domain.

> 
> Thanks,
> James
> 
-- 
WBR, i.m.chubin


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

> Please, show the config file of the domain.
> 
> 
As I said, I''m new to xen. I''ve been calling it a guest OS,
which from
what I''ve read is not really correct. My apologies....

Anyway, I''m trying to create the domain for the "captured"
centos.img
using the Virtual Machine Manager/Create Virtual Machine. When I click
the last "OK" is when I get the error. So it never creates a config
file
for the domain. 

Should I create the domain a different way (don''t know another way!) or
manually create the config file based on another domain that is already
working?

under /etc/xen/vm I only see my other domain in there, not the new one
I''m trying to create.

Thanks,
James


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

On Wed, 2008-04-02 at 14:27 -0400, James Pifer wrote:
> Anyway, I''m trying to create the domain for the
"captured" centos.img
> using the Virtual Machine Manager/Create Virtual Machine. When I click
> the last "OK" is when I get the error. So it never creates a
config file
> for the domain. 
When you got this error, don''t press "OK". Go to your
/etc/xen
directory, look for one new file (ls -ltr /etc/xen) and show us.

-- 
Tiago Cruz
http://everlinux.com
Linux User #282636



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

On Wed, 2008-04-02 at 15:33 -0300, Tiago Cruz wrote:
> On Wed, 2008-04-02 at 14:27 -0400, James Pifer wrote:
> > Anyway, I''m trying to create the domain for the
"captured" centos.img
> > using the Virtual Machine Manager/Create Virtual Machine. When I click
> > the last "OK" is when I get the error. So it never creates a
config file
> > for the domain. 
> 
> When you got this error, don''t press "OK". Go to your
/etc/xen
> directory, look for one new file (ls -ltr /etc/xen) and show us.
> 
Nothing new in /etc/xen or /etc/xen/vm. I did not press ok on the error,
it''s still on the screen. It doesn''t create the config file. 

Screenshot of error here:
http://www.obrien-pifer.com/screenshot.jpg

James


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

On Wed, 2008-04-02 at 14:59 -0400, James Pifer wrote:
> Screenshot of error here:
> http://www.obrien-pifer.com/screenshot.jpg
Maybe I can be wrong, but if you would like to virtualize one Red Hat
v4, you don''t can use "Paravirtualization". You must to use
"Full-Virtualization".

If you guest OS (domU) was one RHEL5, you will can use "ParaVirt"

-- 
Tiago Cruz
http://everlinux.com
Linux User #282636



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

Tiago Cruz wrote:
> On Wed, 2008-04-02 at 14:59 -0400, James Pifer wrote:
> 
> > Screenshot of error here:
> > http://www.obrien-pifer.com/screenshot.jpg
> 
> Maybe I can be wrong, but if you would like to virtualize one Red Hat
> v4, you don''t can use "Paravirtualization". You must to
use
> "Full-Virtualization".
> 
> If you guest OS (domU) was one RHEL5, you will can use "ParaVirt"
RHEL/CentOS 4.6+ has support for xen PV kernels

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

On Wed, 2008-04-02 at 16:09 -0300, Tiago Cruz wrote:
> On Wed, 2008-04-02 at 14:59 -0400, James Pifer wrote:
> 
> > Screenshot of error here:
> > http://www.obrien-pifer.com/screenshot.jpg
> 
> Maybe I can be wrong, but if you would like to virtualize one Red Hat
> v4, you don''t can use "Paravirtualization". You must to
use
> "Full-Virtualization".
> 
> If you guest OS (domU) was one RHEL5, you will can use "ParaVirt"
> 
Okay, tried again with "Full-Virtualization". Now it tries to boot but
gives an error about missing OS. :-( At this point I was just trying to
make it work on principal and futrue reference. Maybe I''ll just do as
Ross suggested and build a new domain and bring the configs over. 

This has probably been covered a million times, but... what is the
preferred method, "Paravirtualization" or
"Full-Virtualization"? I''m
guessing Full-Virtualization because it''s using VT enabled in the
processor.

Thanks to all. 

James


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

James,

On Wed, 2008-04-02 at 15:36 -0400, James Pifer wrote:
> This has probably been covered a million times, but... what is the
> preferred method, "Paravirtualization" or
"Full-Virtualization"? I''m
> guessing Full-Virtualization because it''s using VT enabled in the
> processor.
Para-virt it''s faster ("Lightweight method of virtualizing
machines",
following the help of virt-manager) but only works with one modified
version of kernel.

Full-virt it''s slower, but works with no kernel modification (ex: older
linux versions, windows and others).

-- 
Tiago Cruz
http://everlinux.com
Linux User #282636



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

James Pifer wrote:
> On Wed, 2008-04-02 at 16:09 -0300, Tiago Cruz wrote:
> > On Wed, 2008-04-02 at 14:59 -0400, James Pifer wrote:
> > 
> > > Screenshot of error here:
> > > http://www.obrien-pifer.com/screenshot.jpg
> > 
> > Maybe I can be wrong, but if you would like to virtualize one Red Hat
> > v4, you don''t can use "Paravirtualization". You
must to use
> > "Full-Virtualization".
> > 
> > If you guest OS (domU) was one RHEL5, you will can use
"ParaVirt"
> > 
> 
> Okay, tried again with "Full-Virtualization". Now it tries to
boot but
> gives an error about missing OS. :-( At this point I was just trying to
> make it work on principal and futrue reference. Maybe I''ll just do
as
> Ross suggested and build a new domain and bring the configs over. 
Well, your almost there, but you need to boot up an OS recovery iso
in the HVM and reinstall grub as the primary disk will now be
''hda''
instead of whatever the other server''s was, probably
''sda'', or
''md0'' if it had software raid.

Once you have it running install the xen PV kernel, use disklabels
for non-LVM volumes in your fstab and grub and then shut it down
and convert the config to PVM. Boot with ''xm create -c
<config''
so you see the grub menu and select the xen kernel. Once you
have the PVM running, set it''s kernel as the default in grub.
> This has probably been covered a million times, but... what is the
> preferred method, "Paravirtualization" or
"Full-Virtualization"? I''m
> guessing Full-Virtualization because it''s using VT enabled in the
> processor.
The preferred method is whether you need PV performance. For Linux
based domains I highly recommend PV as you''ll get enhanced CPU
performance and 1Gbe networking instead of 100Mbe networking.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

> Well, your almost there, but you need to boot up an OS recovery iso
> in the HVM and reinstall grub as the primary disk will now be
''hda''
> instead of whatever the other server''s was, probably
''sda'', or
> ''md0'' if it had software raid.
> 
> Once you have it running install the xen PV kernel, use disklabels
> for non-LVM volumes in your fstab and grub and then shut it down
> and convert the config to PVM. Boot with ''xm create -c
<config''
> so you see the grub menu and select the xen kernel. Once you
> have the PVM running, set it''s kernel as the default in grub.
I tried to do that. Since it''s full virt does an actual CD need to be
inserted in the box? I tried adding a virtual CDROM using the ISO, but
it will not boot off of it. 

On the screen I have
ata0 master is the qemu hard disk.
ata0 slave is Unknown device.

I installed a xen PV kernel on the live server before grabbing it. 

thanks,
James


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

James Pifer wrote:
> 
> > Well, your almost there, but you need to boot up an OS recovery iso
> > in the HVM and reinstall grub as the primary disk will now be
''hda''
> > instead of whatever the other server''s was, probably
''sda'', or
> > ''md0'' if it had software raid.
> > 
> > Once you have it running install the xen PV kernel, use disklabels
> > for non-LVM volumes in your fstab and grub and then shut it down
> > and convert the config to PVM. Boot with ''xm create -c
<config''
> > so you see the grub menu and select the xen kernel. Once you
> > have the PVM running, set it''s kernel as the default in grub.
> 
> I tried to do that. Since it''s full virt does an actual CD need to
be
> inserted in the box? I tried adding a virtual CDROM using the ISO, but
> it will not boot off of it. 
Best thing to do is write up a command-line config for your little
domain. Try this to get started (put your specifics where needed):

kernel = "/usr/lib/xen/boot/hvmloader"
builder=''hvm''
memory = 512
name = "MyDomain"
guest_os_type = ''linux''
vcpus=1
vif = [ ''type=ioemu, bridge=xenbr0'' ]
disk = [ ''phy:/dev/vg0/lv3,hda,w'',
''file:/tmp/rescue.iso,hdc,r'' ]
boot="dcna"
vnc=1
vncconsole=1
vncpasswd=''''
serial=''pty''

This should boot into d=cdrom first, then c=hard disk second, followed
by n=network (PXE) third, followed by a=floppy last.
> On the screen I have
> ata0 master is the qemu hard disk.
> ata0 slave is Unknown device.
Not sure which screen you are talking about here, still using
virt-manager? Drop it and start writing config files.

The options are documented in the examples in /etc/xen directory.
> I installed a xen PV kernel on the live server before grabbing it. 
Good fore thought, that''ll save a step later.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.


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

I''m coming in late to the "party" and haven''t gone
through the entire
thread.  Couple things I did notice so I''m going to base my answers on
the following assumptions:

1. You''re attempting full virtualization.

2. Your guest domain configuration file is in Python format located in
/etc/xen.

If the above two are correct:

1. Open up the Python config file for your CentOS domain.

2. Locate the line starting with "disk ="

3. Within the last double-quote, but before closing bracket, put in the
following immediately after the last double-quote:

, "file:/directory/with/cdrom_dvd_image.iso,hdc:cdrom,r"

So it should look similar to this:

disk = [ "file:/var/lib/xen/images/centos.img,hda,w",
"file:/directory/with/cdrom_dvd_image.iso,hdc:cdrom,r" ]

4. While still the Python domain Xen configuration file, find the line
which starts with "boot =".  It''s probably set to boot =
"c".

5. Change the boot sequence to boot = "dc" or "boot =
"d".

6. Once you''re done with the CD/DVD, you can revert it back.

Good luck!

James Pifer wrote:

I tried to do that. Since it''s full virt does an actual CD need to be
inserted in the box? I tried adding a virtual CDROM using the ISO, but
it will not boot off of it. 

On the screen I have
ata0 master is the qemu hard disk.
ata0 slave is Unknown device.

I installed a xen PV kernel on the live server before grabbing it. 

thanks,
James

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

-- 
Daniel Kao
Übermind, Inc.
dkao@ubermind.com
Seattle, WA, U.S.A.


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

> Not sure which screen you are talking about here, still using
> virt-manager? Drop it and start writing config files.
> 
> The options are documented in the examples in /etc/xen directory.

Trying this. Should I still use virt-manager for starting and stopping
them? If so, how do I get it to re-read the new config file? If not, how
do you start them without virt-manager?

Thanks,
James


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

You will start it from the command line with ''xm create <config
file>''.

Oh and I made a mistake in my example config earlier, where it specified the iso
for the cdrom it should be ''file:/path_to_iso,hdc:cdrom,r'', I
left out the :cdrom for the device type.

-Ross


----- Original Message -----
From: James Pifer <jep@obrien-pifer.com>
To: Ross S. W. Walker
Cc: Xen List <xen-users@lists.xensource.com>
Sent: Wed Apr 02 18:02:20 2008
Subject: RE: [Xen-users] Convert CentOS system to Xen image?
> Not sure which screen you are talking about here, still using
> virt-manager? Drop it and start writing config files.
> 
> The options are documented in the examples in /etc/xen directory.

Trying this. Should I still use virt-manager for starting and stopping
them? If so, how do I get it to re-read the new config file? If not, how
do you start them without virt-manager?

Thanks,
James


______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.



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

On Mi, Apr 02, 2008 at 02:59:11 -0400, James Pifer
wrote:
> On Wed, 2008-04-02 at 15:33 -0300, Tiago Cruz wrote:
> > On Wed, 2008-04-02 at 14:27 -0400, James Pifer wrote:
> > > Anyway, I''m trying to create the domain for the
"captured" centos.img
> > > using the Virtual Machine Manager/Create Virtual Machine. When I
click
> > > the last "OK" is when I get the error. So it never
creates a config file
> > > for the domain. 
> > 
> > When you got this error, don''t press "OK". Go to
your /etc/xen
> > directory, look for one new file (ls -ltr /etc/xen) and show us.
> > 
> 
> Nothing new in /etc/xen or /etc/xen/vm. I did not press ok on the error,
> it''s still on the screen. It doesn''t create the config
file.
> 
> Screenshot of error here:
> http://www.obrien-pifer.com/screenshot.jpg

Hello James, 


I''m sorry for delay, I was away from the network.



Let us check image you have grabbed.
Is it ok or no.


Question #1.

If you have qemu installed, 
just try to run

qemu -hda centos.img

and say (or better show)
what do you see.


Question #2.

(if QEMU is not installed and you can''t install it)

What does the command chow?

# file centos.img



Question #3.


# lsmod | grep -q ^loop || modprobe loop
# losetup /dev/loop1 centos.img  
  (if /dev/loop1 is busy, try /dev/loop2, /dev/loop3 and so on
  you can get know what loop devices are busy with the command
  losetup -a
  if device is not listed, it is not busy and you can use it)

# fdisk -l /dev/loop1

And show what you have got.










> 
> James
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
-- 
WBR, i.m.chubin


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

> I''m sorry for delay, I was away from the network.
> 
NP. Beggers can''t be choosers. :-)
> 
> 
> Let us check image you have grabbed.
> Is it ok or no.
> 
> 
> Question #1.
> 
> If you have qemu installed, 
> just try to run
> 
> qemu -hda centos.img
Not installed. 
> 
> and say (or better show)
> what do you see.
> 
> 
> Question #2.
> 
> (if QEMU is not installed and you can''t install it)
> 
> What does the command chow?
> 
> # file centos.img
> 
This looks good I think.

# file centos.img
centos.img: x86 boot sector, Microsoft Windows XP mbr,Serial 0xb5058

> 
> Question #3.
> 
> 
> # lsmod | grep -q ^loop || modprobe loop
> # losetup /dev/loop1 centos.img  
>   (if /dev/loop1 is busy, try /dev/loop2, /dev/loop3 and so on
>   you can get know what loop devices are busy with the command
>   losetup -a
>   if device is not listed, it is not busy and you can use it)
> 
> # fdisk -l /dev/loop1
> 
> And show what you have got.
This can''t be good.....

# lsmod | grep -q ^loop || modprobe lo
# losetup /dev/loop1 centos.img
# fdisk -l /dev/loop1

Disk /dev/loop1: 3845 MB, 3845230592 bytes
255 heads, 63 sectors/track, 467 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/loop1 doesn''t contain a valid partition table

Thanks,
James



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

...
> 
> This looks good I think.
> 
> # file centos.img
> centos.img: x86 boot sector, Microsoft Windows XP mbr,Serial 0xb5058
Can you run 

file -s /dev/ida/c0d0

on the host where from
you have copied the image?



> 
> 
> > 
> > Question #3.
> > 
> > 
> > # lsmod | grep -q ^loop || modprobe loop
> > # losetup /dev/loop1 centos.img  
> >   (if /dev/loop1 is busy, try /dev/loop2, /dev/loop3 and so on
> >   you can get know what loop devices are busy with the command
> >   losetup -a
> >   if device is not listed, it is not busy and you can use it)
> > 
> > # fdisk -l /dev/loop1
> > 
> > And show what you have got.
> 
> This can''t be good.....
> 
> # lsmod | grep -q ^loop || modprobe lo
> # losetup /dev/loop1 centos.img
> # fdisk -l /dev/loop1
> 
> Disk /dev/loop1: 3845 MB, 3845230592 bytes
> 255 heads, 63 sectors/track, 467 cylinders
> Units = cylinders of 16065 * 512 = 8225280 bytes
> 
> Disk /dev/loop1 doesn''t contain a valid partition table
What shows fdisk -l /dev/ida/c0d0 on the source host?


Size=4G <- is it ok?
You have said that your HDD is 8GB size

> 
> Thanks,
> James
> 
> 
-- 
WBR, i.m.chubin


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

> Can you run 
> 
> file -s /dev/ida/c0d0
> 
> on the host where from
> you have copied the image?
> 
# file -s /dev/ida/c0d0
/dev/ida/c0d0: x86 boot sector

> Size=4G <- is it ok?
> You have said that your HDD is 8GB size

No, actually the used space is a little over 8GB. The LV is about 18GB.

James


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

On Mi, Apr 02, 2008 at 09:19:53 -0400, James Pifer
wrote:
> > Can you run 
> > 
> > file -s /dev/ida/c0d0
> > 
> > on the host where from
> > you have copied the image?
> > 
> 
> # file -s /dev/ida/c0d0
> /dev/ida/c0d0: x86 boot sector
> 
> 
> > Size=4G <- is it ok?
> > You have said that your HDD is 8GB size
> 
> 
> No, actually the used space is a little over 8GB. The LV is about 18GB.

fdisk output shat you have showed before
says, that your image is about 4G.

It means that you haven''t copied entire disk.


When you do raw data copying (using cat or dd),
result file will have size of the source disk (18G in your case,
or even more if your disk has greater size; and 18G is LV, but not
the disk, size).


So, it''s better, may be, to do not raw copying in your case, 
but files copying.


If you have LVM installed on target system, you may create
new LV, format it and mount.


E.g.:


lvcreate -L 8G -n centos /dev/VG
mkfs.ext3 /dev/VG/centos
mount /dev/VG/centos /mnt/


If you have not, than you regular files:

dd if=/dev/zero of=centos-new.img count=$[8*1024*1024] bs=1k
mkfs.ext3 centos-new.img  # (press y here)
mount -o loop centos-new.img /mnt/


After that
you may start to copy files:


target#   rsync -a --exclude=/sys --exclude=/proc --exclude=/dev source:/ /mnt/

When copying will be finished:

target#  mkdir -p /mnt/{sys,proc,dev}
target#  cp -a /lib/modules/VERSION_OF_THE_PV_KERNEL_HERE /mnt/lib/
target#  cat <<EOF > /mnt/etc/fstab
proc                        /proc               proc    defaults 0       0
/dev/hda1                   /                   ext3    defaults 0       0
EOF

After that 

umount /mnt


And create config file for new domain.


This is a PV domain, not an HVM!
So you need new config file.



> 
> James
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
-- 
WBR, i.m.chubin


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

Hello guys,

Just my 2 cents:

On Thu, 2008-04-03 at 11:30 +0300, Igor Chubin wrote:
> So, it''s better, may be, to do not raw copying in your case, 
> but files copying.
James, I''ve had did a lot of migrations using one machine real into one
virtual (also called "p2v") "at hot" -- without shutdown the
machine.

If you can turn off the physical machine, please, considere to use one
tool to automatize your migration, like this one:
http://et.redhat.com/~rjones/virt-p2v/


Whatever, I know this ways to do this work "by hand":

1-) Cloning the device using dd+nc
2-) Coping all "/" using rsync

For this second way, I recommend this read:
http://wiki.xensource.com/xenwiki/XenManualPtoVProcess

And I can say one example of "rsync" with a lot of parameters:

rsync -vaH --numeric-ids --stats --progress --exclude "/mnt/*" \
--exclude "/proc/*" --exclude "/sys/*" --exclude
"/tmp/*" \
--exclude "/var/tmp/*" --exclude "/var/run/*.pid" \
--exclude "/var/run/dbus/system_bus_socket"\
rsync://192.168.1.123/p2v/ /mnt/vmname/

http://wiki.xen-br.org/index.php?title=P2v-howto

But, in both wais, you can have boot problems with LiLo/Grub, and with
different devices (ex: /dev/ida/c0d0p1 -> /dev/hda1).

So, i think that the easier way to do this, its the follow:

- Install one new CentOS inside Xen, using Virt-Manager with sufficient
space to do the migration;
- Boot this one, test, and lean how to start and stop this virtual
machine;
- Backup this machine;
- Just "rsync" all "/" from the physical one for inside of
"/" of
virtual one;
- Restore your backup from /etc/hosts, /etc/fstab, /etc/mtab/
ifcfg-eth-x, lilo.conf, menu.lst and be happy :-)

May the Force be with you!

-- 
Tiago Cruz
http://everlinux.com
Linux User #282636



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

> So, i think that the easier way to do this, its the follow:
> 
> - Install one new CentOS inside Xen, using Virt-Manager with sufficient
> space to do the migration;
> - Boot this one, test, and lean how to start and stop this virtual
> machine;
> - Backup this machine;
> - Just "rsync" all "/" from the physical one for inside
of "/" of
> virtual one;
> - Restore your backup from /etc/hosts, /etc/fstab, /etc/mtab/
> ifcfg-eth-x, lilo.conf, menu.lst and be happy :-)
> 
> May the Force be with you!
> 
Igor, I followed your last suggestion and now I''m back to "A valid
boot
sector was not found. The installation may have failed." No config file
has been written yet either. (yes, using virt-manager)

All the steps you wrote went without a hitch and I verified fstab was
updated correctly. 

Tiago, I''ve tried installing CentOS 4.5, 4.6, and 5.1 using
virt-manager, but none of them see a disk to install to once the CentOS
install process starts. Maybe there''s a special way to run these
installs. I''ll have to google for that. SLES10.1 installs without doing
anything special. 

Thanks,
James


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

On Do, Apr 03, 2008 at 10:11:04 -0400, James Pifer
wrote:
> > So, i think that the easier way to do this, its the follow:
> > 
> > - Install one new CentOS inside Xen, using Virt-Manager with
sufficient
> > space to do the migration;
> > - Boot this one, test, and lean how to start and stop this virtual
> > machine;
> > - Backup this machine;
> > - Just "rsync" all "/" from the physical one for
inside of "/" of
> > virtual one;
> > - Restore your backup from /etc/hosts, /etc/fstab, /etc/mtab/
> > ifcfg-eth-x, lilo.conf, menu.lst and be happy :-)
> > 
> > May the Force be with you!
> > 
> 
> Igor, I followed your last suggestion and now I''m back to "A
valid boot
> sector was not found. The installation may have failed." No config
file
> has been written yet either. (yes, using virt-manager)
> 
> All the steps you wrote went without a hitch and I verified fstab was
> updated correctly. 

Please, show the config file you use
to start the virtual machine.

It should be situated in the /etc/xen/ directory.

Also, try to use command lines tools like xm
and after success with them you may switch 
to virt-manager




> 
> Tiago, I''ve tried installing CentOS 4.5, 4.6, and 5.1 using
> virt-manager, but none of them see a disk to install to once the CentOS
> install process starts. Maybe there''s a special way to run these
> installs. I''ll have to google for that. SLES10.1 installs without
doing
> anything special. 
> 
> Thanks,
> James
> 
-- 
WBR, i.m.chubin


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

On Thu, 2008-04-03 at 10:11 -0400, James Pifer wrote:
> Tiago, I''ve tried installing CentOS 4.5, 4.6, and 5.1 using
> virt-manager, but none of them see a disk to install to once the CentOS
> install process starts. Maybe there''s a special way to run these
> installs.
I''m just installing new RHEL5 machines doing this. Takes ~4 minutes.

# virt-install \
--name=myvm \
--ram=1000 \
--file=/dev/Vol_LVM/vm05 \
--network=bridge:virbr0 \
--vnc \
--paravirt \
--location=http://server.com/redhat5as_install/  \
-x ks=http://server.com/redhat5as_install/ks.cfg

Where:
- /dev/Vol_LVM/vm05 its one LVM device with 20 GB
- ks.cfg:

install
url --url http://server.com/redhat5as_install/
lang en_US.UTF-8
keyboard br-abnt2
network --device eth0 --bootproto dhcp --hostname xen.com.br
rootpw --iscrypted (#hash of password#)
firewall --disable
authconfig --enableshadow --enablemd5
selinux --disable
timezone --utc America/Sao_Paulo
bootloader --location=mbr --driveorder=xvda
clearpart --linux
part / --fstype ext3 --size=100 --grow
part swap --size=3000 --asprimary

%packages
%packages
@admin-tools
@base
@core
@development-libs
@development-tools
@dialup
@editors
@printing
@system-tools
@text-internet


> 
-- 
Tiago Cruz
http://everlinux.com
Linux User #282636



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

> Please, show the config file you use
> to start the virtual machine.
> 
> It should be situated in the /etc/xen/ directory.
> 
> Also, try to use command lines tools like xm
> and after success with them you may switch 
> to virt-manager
> 
> 
I looked back through this thread and don''t see anywhere that
I''ve said
what dom0 is running. It''s running on 64 bit SLES 10.1. 

I don''t understand what to specify on the kernel line. Right now I have
this:

kernel = "/usr/lib/xen/boot/hvmloader"
builder=''linux''
name = "pcalakwww01"
memory = "1024"
disk = [ ''file:/etc/xen/images/pcalakwww01/centos.img,xvda,w''
]
vif = [ ''bridge=xenbr1'', ]
vcpus=1
on_reboot = ''restart''
on_crash = ''destroy''
vfb=["type=vnc,vncunused=1"]

When I try to run xm create pcalakwww01 I get:
Using config file "./pcalakwww01".
Error: (2, ''Invalid kernel'', "elf_xen_note_check: ERROR:
Not a Xen-ELF image: No ELF notes or ''__xen_guest'' section
found.\n")

Also, once I run this, how do you convert it or get it into virt-manager? I see
it in there, but if I do shutdown it goes away.

Thanks,
James


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

Hello again guys,

This thread was very interesting for me, so, I''ve decided do some
tests,
and I would like to share with the list (and hope to kill some
doubts...)

So, I have running on my desktop (Dell Optiplex 745) one Ubuntu Hardy,
on my sda2, so:

- dd+netcat to copy my device into my XenServer running RHEL5;
- create one new LVM file, and particione it (/ and swap);
- mount the image and copy the files to LVM device;
- adjust files like fstab and mtab (/ its now sda1, not more sda2)

and after, I''ve used the Igor''s idea:

On Thu, 2008-04-03 at 02:00 +0300, Igor Chubin wrote:
> just try to run
> 
> qemu -hda centos.img
> 
> and say (or better show)
I never was think on this before, great idea! Before kill myself using
Xen, I''ve used the old and good qemu to fix my image (reinstall grub
and
etc) :-D

After this, I''ve just put my working qemu''s image inside Xen,
like this:

name = "teste"
uuid = "da7371c1deaf8280213a0534b43da4ca"
maxmem = 500
memory = 300
vcpus = 1
builder = "hvm"
kernel = "/usr/lib/xen/boot/hvmloader"
boot = "c"
pae = 1
acpi = 1
apic = 1
on_poweroff = "destroy"
on_reboot = "restart"
on_crash = "restart"
device_model = "/usr/lib64/xen/bin/qemu-dm"
sdl = 0
vnc = 1
vncunused = 1
keymap = "en-us"
disk = [ "phy:/dev/Vol_LVM/teste,hda,w",
"file:/iso/ubuntu-7.10-desktop-i386.iso,hdc:cdrom,r" ]
vif = [ "mac=00:16:3e:1f:86:70,bridge=virbr0,type=ioemu" ]
serial = "pty"


And it''s working like a charm, but:

1-) Should I use hda or sda on xen config file? And on domU (grub,
fstab)... must be hda or sda?

2-) I''m getting a lot of errors like this on my dmesg/messages... what
can I do to fix this? 

Apr 11 17:04:46 tuxkiller kernel: [ 6847.639957] ata2: DRQ=1 with device error,
dev_stat 0x49
Apr 11 17:04:46 tuxkiller kernel: [ 6847.640405]          res
41/50:03:00:00:20/00:00:00:00:00/a0 Emask 0x3 (HSM violation)
Apr 11 17:04:46 tuxkiller kernel: [ 6847.641026] ata2: soft resetting port
Apr 11 17:04:46 tuxkiller kernel: [ 6848.106528] ata2.00: configured for MWDMA2
Apr 11 17:04:46 tuxkiller kernel: [ 6848.106540] ata2: EH complete

My boot dmesg show me this:

[    9.399068] eth0: RealTek RTL8139 at 0xd380e000, 00:16:3e:1f:86:70, IRQ 16
[    9.399542] eth0:  Identified 8139 chip type ''RTL-8139''
[    9.403628] 8139cp: 10/100 PCI Ethernet driver v1.3 (Mar 22, 2004)
[    9.427008] sd 0:0:0:0: [sda] 229015552 512-byte hardware sectors (117256 MB)
[    9.427604] sd 0:0:0:0: [sda] Write Protect is off
[    9.428137] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[    9.428159] sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled,
doesn''t support DPO or FUA
[    9.430447] sd 0:0:0:0: [sda] 229015552 512-byte hardware sectors (117256 MB)
[    9.431032] sd 0:0:0:0: [sda] Write Protect is off
[    9.431445] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[    9.431457] sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled,
doesn''t support DPO or FUA
[    9.434390]  sda: sda1
[    9.436832] sd 0:0:0:0: [sda] Attached SCSI disk
[    9.439554] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
[    9.440016] ata2.00: (BMDMA stat 0x5)
[    9.440392] ata2.00: cmd a0/01:00:00:00:00/00:00:00:00:00/a0 tag 0 cdb 0x5a
data 128 in
[    9.440394]          res 48/00:02:00:00:00/00:00:00:00:00/a0 Emask 0x2 (HSM
violation)
[    9.442395] ata2: soft resetting port
[    9.448559] sd 0:0:0:0: Attached scsi generic sg0 type 0
[    9.449085] sr 1:0:0:0: Attached scsi generic sg1 type 5
[    9.913637] ata2.00: configured for MWDMA2
[    9.914076] ata2: EH complete
[    9.914748] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
[    9.917331] ata2.00: (BMDMA stat 0x5)
[    9.917777] ata2.00: cmd a0/01:00:00:00:00/00:00:00:00:00/a0 tag 0 cdb 0x5a
data 128 in
[    9.917778]          res 48/01:02:00:00:00/00:00:00:00:00/a0 Emask 0x2 (HSM
violation)
[    9.919393] ata2: soft resetting port
[   10.388618] ata2.00: configured for MWDMA2
[   10.389032] ata2: EH complete
[   10.389605] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
[   10.390070] ata2.00: (BMDMA stat 0x5)
[   10.392294] ata2.00: cmd a0/01:00:00:00:00/00:00:00:00:00/a0 tag 0 cdb 0x5a
data 128 in
[   10.392296]          res 48/01:02:00:00:00/00:00:00:00:00/a0 Emask 0x2 (HSM
violation)
[   10.393724] ata2: soft resetting port
[   10.859690] ata2.00: configured for MWDMA2
[   10.860317] ata2: EH complete
[   10.861011] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
[   10.863270] ata2.00: (BMDMA stat 0x5)
[   10.863658] ata2.00: cmd a0/01:00:00:00:00/00:00:00:00:00/a0 tag 0 cdb 0x5a
data 128 in
[   10.863660]          res 48/01:02:00:00:00/00:00:00:00:00/a0 Emask 0x2 (HSM
violation)
[   10.866820] ata2: soft resetting port
[   11.334555] ata2.00: configured for MWDMA2
[   11.335059] ata2: EH complete
[   11.335515] sr0: scsi-1 drive
[   11.335969] Uniform CD-ROM driver Revision: 3.20
[   11.338453] sr 1:0:0:0: Attached scsi CD-ROM sr0
[   14.786436] kjournald starting.  Commit interval 5 seconds
[   14.787029] EXT3-fs: mounted filesystem with ordered data mode.
[   19.115422] udev: renamed network interface eth0 to eth1
[   21.326355] input: PC Speaker as /class/input/input2
[   21.420419] input: ImExPS/2 Generic Explorer Mouse as /class/input/input3
[   21.829974] lp: driver loaded but no devices found
[   22.256104] EXT3 FS on sda1, internal journal
[   23.581239] Adding 1023992k swap on /swap.  Priority:-1 extents:259
across:1053904k
[   23.754162] AppArmor: Unable to register AppArmor
[   25.654011] No dock devices found.
[   27.890021] ata2: DRQ=1 with device error, dev_stat 0x49
[   27.891842] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
[   27.892709] ata2.00: (BMDMA stat 0x5)
[   27.893998] ata2.00: cmd a0/01:00:00:00:00/00:00:00:00:00/a0 tag 0 cdb 0x46
data 32 in
[   27.894000]          res 41/50:03:00:00:00/00:00:00:00:00/a0 Emask 0x3 (HSM
violation)
[   27.894906] ata2: soft resetting port
[   28.361448] ata2.00: configured for MWDMA2
[   28.361471] ata2: EH complete


# lspci 
00:00.0 Host bridge: Intel Corporation 440FX - 82441FX PMC [Natoma] (rev 02)
00:01.0 ISA bridge: Intel Corporation 82371SB PIIX3 ISA [Natoma/Triton II]
00:01.1 IDE interface: Intel Corporation 82371SB PIIX3 IDE [Natoma/Triton II]
00:01.2 Bridge: Intel Corporation 82371AB/EB/MB PIIX4 ACPI (rev 01)
00:02.0 VGA compatible controller: Cirrus Logic GD 5446
00:03.0 Class ff80: XenSource, Inc. Xen Platform Device (rev 01)
00:04.0 Ethernet controller: Realtek Semiconductor Co., Ltd.
RTL-8139/8139C/8139C+ (rev 20)


Many thanks!

-- 
Tiago Cruz
http://everlinux.com
Linux User #282636



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

On Fri, 2008-04-11 at 14:09 -0300, Tiago Cruz wrote:

I''ve removed the "cdrom" device
> disk = [ "phy:/dev/Vol_LVM/teste,hda,w",
"file:/iso/ubuntu-7.10-desktop-i386.iso,hdc:cdrom,r" ]
And this erros stopped :-)

> [   27.893998] ata2.00: cmd a0/01:00:00:00:00/00:00:00:00:00/a0 tag 0 cdb
0x46 data 32 in
> [   27.894000]          res 41/50:03:00:00:00/00:00:00:00:00/a0 Emask 0x3
(HSM violation)
> [   27.894906] ata2: soft resetting port
> [   28.361448] ata2.00: configured for MWDMA2
> [   28.361471] ata2: EH complete



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