Mailing List Archive

On Tue, Jul 05, 2005 at 05:54:32PM +0200, Miguel Gómez wrote:
> Hello people,
> I have been testing Xen and I want to migrate a Domain from a machine to
> another. It will not even be a live migration, it will be a normal one. But
> when I try to start xfrd, it says that it doesn't find a library, the
> libcurl one:
>
> XENslave:/# xfrd start
> xfrd: error while loading shared libraries: libcurl.so.2: cannot open shared
> object file: No such file or directory
>
> And it is true, it doesn't exist, if I do:
>
> XENslave:/# ldd /usr/sbin/xfrd
> libxc.so.2.0 => /usr/lib/libxc.so.2.0 (0x4001b000)
> libxutil.so.2.0 => /usr/lib/libxutil.so.2.0 (0x4002a000)
> libz.so.1 => /usr/lib/libz.so.1 (0x4003b000)
> libcurl.so.2 => not found
> libssl.so.4 => not found
> libcrypto.so.4 => not found
> libgssapi_krb5.so.2 => not found
> libkrb5.so.3 => not found
> libcom_err.so.2 => /lib/libcom_err.so.2 (0x4004e000)
> libk5crypto.so.3 => not found
> libresolv.so.2 => /lib/libresolv.so.2 (0x40051000)
> libdl.so.2 => /lib/libdl.so.2 (0x40063000)
> libc.so.6 => /lib/libc.so.6 (0x40066000)
> /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
>
> So, I don't really know what to do, I have done a libcurl installation but
> it is libcurl3, not libcurl2. There's a thread in this mailing list:
>
> http://lists.xensource.com/archives/html/xen-users/2005-07/msg00003.html
>
> But it doesn't really help me. I have Debian Sarge installed, and XEN runs
> normally. But I've got these problems when I try to run xfrd. How can I
> resolve it? Symbolic link to libcurl.so.3? Should I try to get back to
> libcurl2?


Naw, you don't make symlinks - you can't know if blablub() in libcurl2 does
even nearly the same as in libcurl3.


You can install both, no?

> _______________________________________________
> 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

On Tue, 5 Jul 2005, [ISO-8859-1] Miguel Gómez wrote:

> Hello people,
> I have been testing Xen and I want to migrate a Domain from a machine to
> another. It will not even be a live migration, it will be a normal one. But
> when I try to start xfrd, it says that it doesn't find a library, the
> libcurl one:
>
> XENslave:/# xfrd start
> xfrd: error while loading shared libraries: libcurl.so.2: cannot open shared
> object file: No such file or directory
>
> And it is true, it doesn't exist, if I do:
>
> XENslave:/# ldd /usr/sbin/xfrd
> libxc.so.2.0 => /usr/lib/libxc.so.2.0 (0x4001b000)
> libxutil.so.2.0 => /usr/lib/libxutil.so.2.0 (0x4002a000)
> libz.so.1 => /usr/lib/libz.so.1 (0x4003b000)
> libcurl.so.2 => not found
> libssl.so.4 => not found
> libcrypto.so.4 => not found
> libgssapi_krb5.so.2 => not found
> libkrb5.so.3 => not found
> libcom_err.so.2 => /lib/libcom_err.so.2 (0x4004e000)
> libk5crypto.so.3 => not found
> libresolv.so.2 => /lib/libresolv.so.2 (0x40051000)
> libdl.so.2 => /lib/libdl.so.2 (0x40063000)
> libc.so.6 => /lib/libc.so.6 (0x40066000)
> /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
>
> So, I don't really know what to do, I have done a libcurl installation but
> it is libcurl3, not libcurl2. There's a thread in this mailing list:
>
> http://lists.xensource.com/archives/html/xen-users/2005-07/msg00003.html
>
> But it doesn't really help me. I have Debian Sarge installed, and XEN runs
> normally. But I've got these problems when I try to run xfrd. How can I
> resolve it? Symbolic link to libcurl.so.3? Should I try to get back to
> libcurl2?
>

Symbolic link from libcurl.so.3 to libcurl.so.2 works for me. see the faq
of xen. This is well known problem in fedora, but not sure it works for
debian.

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

On Tue, 5 Jul 2005, Sascha Retzki wrote:

> On Tue, Jul 05, 2005 at 05:54:32PM +0200, Miguel Gómez wrote:
> > Hello people,
> > I have been testing Xen and I want to migrate a Domain from a machine to
> > another. It will not even be a live migration, it will be a normal one. But
> > when I try to start xfrd, it says that it doesn't find a library, the
> > libcurl one:
> >
> > XENslave:/# xfrd start
> > xfrd: error while loading shared libraries: libcurl.so.2: cannot open shared
> > object file: No such file or directory
> >
> > And it is true, it doesn't exist, if I do:
> >
> > XENslave:/# ldd /usr/sbin/xfrd
> > libxc.so.2.0 => /usr/lib/libxc.so.2.0 (0x4001b000)
> > libxutil.so.2.0 => /usr/lib/libxutil.so.2.0 (0x4002a000)
> > libz.so.1 => /usr/lib/libz.so.1 (0x4003b000)
> > libcurl.so.2 => not found
> > libssl.so.4 => not found
> > libcrypto.so.4 => not found
> > libgssapi_krb5.so.2 => not found
> > libkrb5.so.3 => not found
> > libcom_err.so.2 => /lib/libcom_err.so.2 (0x4004e000)
> > libk5crypto.so.3 => not found
> > libresolv.so.2 => /lib/libresolv.so.2 (0x40051000)
> > libdl.so.2 => /lib/libdl.so.2 (0x40063000)
> > libc.so.6 => /lib/libc.so.6 (0x40066000)
> > /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
> >
> > So, I don't really know what to do, I have done a libcurl installation but
> > it is libcurl3, not libcurl2. There's a thread in this mailing list:
> >
> > http://lists.xensource.com/archives/html/xen-users/2005-07/msg00003.html
> >
> > But it doesn't really help me. I have Debian Sarge installed, and XEN runs
> > normally. But I've got these problems when I try to run xfrd. How can I
> > resolve it? Symbolic link to libcurl.so.3? Should I try to get back to
> > libcurl2?
>
>
> Naw, you don't make symlinks - you can't know if blablub() in libcurl2 does
> even nearly the same as in libcurl3.
>
>
> You can install both, no?

Not easily, because libcurl2 was removed from debian sarge.

I would recommend to use xen-testing which works fine with libcurl3.

Or you could try to use libcurl2 from oldstable - no guarantee that this
works (eg from

ftp://ftp.debian.org/debian/pool/main/c/curl/libcurl2_7.9.5-1_i386.deb

).

Rainer Schöpf

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

hello,
Thank you for your responses, I've done what you say, so I have created the
symlinks from libcurl.so.2 to libcurl.so.3 and now I can start xfrd, it
finds all the libraries it needs, and it looks fine to me.

I did that in two machines and I tried to migrate a Domain (ttylinux domain)
from a machine to the other one, but the migration fails. I am not trying to
do a live migration, so I don't need to have a NFS server with the VBD in
it, do I?

The migration fails and I get this error:
Error: errors: transfer daemon (xfrd) error: 1
And I've got stucked in that. Yesterday I found a messaje in Xen-users
mail-list talking about it, but It doesn't get into any conclusion.

xfrd starts, and if I do netstat -na, I find that there's a service
listening at port 8002, is this correct? xfrd has to do that, doesn't it?

Then I execute

#xm migrate Debian1 192.168.0.4 <http://192.168.0.4>

And I get the error. Both of the machines are in the same subnet. Where
should I look for the problem?
Thank you for your support and patiente.

Regards,
Miguel.

That's not the only problem that I get, I can not save and restore domains.
Take a look at these commands:

XENmaster:/xen# xm save ttylinux1 Savettylinux1
XENmaster:/xen# ls
Savettylinux1 vmttylinux1 xmttylinuxNFS_MIGRATE
XENmaster:/xen# xm list
Name Id Mem(MB) CPU State Time(s) Console
Domain-0 0 26 0 r---- 1702.4
XENmaster:/xen# xm restore Savettylinux1
Error: errors: restore failed
XENmaster:/xen# xm list
Name Id Mem(MB) CPU State Time(s) Console
Domain-0 0 26 0 r---- 1703.0
Domain-13 13 16 0 --p-- 0.0

I don't know if this has to do something with the migrate problems. Any
ideas?

Ok, things seem to get better. I was using the pre-built Xen installation.
When I compiled the source distribution, save/restore problem disappeared.
But I can not still migrate domains. When I have a ttylinux domain like
this:

XENslave:/xen# xm list
Name Id Mem(MB) CPU State Time(s) Console
Domain-0 0 26 0 r---- 107.3
ttylinux2 11 7 0 -b--- 0.3 9611

I try to migrate it with:

XENslave:/xen# xm migrate ttylinux2 192.168.50.194 <http://192.168.50.194>

And look what I get in the destiny machine /var/log/xfrd.log:

3211 [INF] XFRD> Forked child pid=3679
3211 [INF] XFRD> Accepted connection from
192.168.50.212:4844<http://192.168.50.212:4844>on 2
3732 [INF] XFRD> Xfr service for 192.168.50.212:4844<http://192.168.50.212:4844>
[DEBUG] Conn_init> flags=1
[DEBUG] Conn_init> write stream...
[DEBUG] stream_init>mode=w flags=1 compress=0
[DEBUG] stream_init> unbuffer...
[DEBUG] stream_init< err=0
[DEBUG] Conn_init> read stream...
[DEBUG] stream_init>mode=r flags=1 compress=0
[DEBUG] stream_init> unbuffer...
[DEBUG] stream_init< err=0
[DEBUG] Conn_sxpr>
(xfr.hello 1 0)[DEBUG] Conn_sxpr< err=0
[DEBUG] Conn_sxpr>
(xfr.xfr 12)[DEBUG] Conn_sxpr< err=0
[1120810614.947681] xc_linux_restore start

xc_linux_restore start
[1120810615.015532] Created domain 9

Created domain 9
(Domain-9 Domain-0)'domain id=9 name=ttylinux2 memory=8 console=9609
image=/boot/vmlinuz-2.6.11-xenU'[1120810
617.262328] Reloading memory pages: 0%
Reloading memory pages: 50% 99%
[1120810627.283337] Received all pages

Received all pages 100%
100%
[1120810627.287122] Memory reloaded.

Memory reloaded.
Decreased reservation by 5 pages
[1120810627.310166] Domain ready to be built.

Domain ready to be built.
[1120810627.310282] Domain ready to be unpaused

Domain ready to be unpaused
[1120810627.312554] DOM=9

DOM=9
[DEBUG] Conn_sxpr>
(xfr.err 0)[DEBUG] Conn_sxpr< err=0
[DEBUG] Conn_sxpr>
(xfr.err 0)[DEBUG] Conn_sxpr< err=0
3732 [INF] XFRD> Transfer complete in 13 seconds
3732 [INF] XFRD> Xfr service err=0




It looks good!!!, but when I do a "xm list" in the destiny machine:
XENmaster:/etc# xm list
Name Id Mem(MB) CPU State Time(s) Console
Domain-0 0 26 0 r---- 76.7

Something is wrong, do you know what I can do? Where can I search for the
problem?

Thank you, everybody, for your help and have a nice weekend!!!!

More information...

The /var/log/xend.log shows me some things:

#################################

[2005-07-08 10:16:55 xend] INFO (XendRoot:112) EVENT>
xend.domain.create['Domain-9', '9']
[2005-07-08 10:16:55 xend] DEBUG (XendDomain:384) domain_configure> id=9
config=['vm', ['name', 'ttylinux2'], ['memory', '8'
[2005-07-08 10:16:55 xend] INFO (console:94) Created console id=13 domain=9
port=9609
[2005-07-08 10:16:55 xend] DEBUG (XendDomainInfo:1130) Creating vbd dom=9
uname=file:/xen/vmttylinux2
[2005-07-08 10:16:55 xend] DEBUG (blkif:63) exception looking up device
number for sda1: [Errno 2] No such file or directory
[2005-07-08 10:16:56 xend] DEBUG (blkif:155) Connecting blkif
<BlkifBackendInterface 9 0>
[2005-07-08 10:16:56 xend] DEBUG (XendDomainInfo:1107) Creating vif dom=9
vif=0 mac=aa:00:00:68:e8:12
[2005-07-08 10:16:56 xend] INFO (XendRoot:112) EVENT>
xend.console.create[13, 9, 9609]
[2005-07-08 10:16:57 xend] INFO (XendRoot:112) EVENT>
xend.domain.create['ttylinux2', '9']
[2005-07-08 10:17:07 xend] DEBUG (XendDomain:244) XendDomain>reap> domain
died name=ttylinux2 id=9
[2005-07-08 10:17:07 xend] INFO (XendDomain:564) Destroying domain:
name=ttylinux2
[2005-07-08 10:17:07 xend] DEBUG (XendDomainInfo:665) Destroying vifs for
domain 9
[2005-07-08 10:17:07 xend] DEBUG (netif:305) Destroying vif domain=9 vif=0
[2005-07-08 10:17:07 xend] DEBUG (XendDomainInfo:674) Destroying vbds for
domain 9
[2005-07-08 10:17:07 xend] DEBUG (blkif:552) Destroying blkif domain=9
[2005-07-08 10:17:07 xend] DEBUG (blkif:408) Destroying vbd domain=9 idx=0
[2005-07-08 10:17:07 xend] DEBUG (XendDomainInfo:634) Closing console,
domain 9
[2005-07-08 10:17:07 xend] DEBUG (XendDomainInfo:622) Closing channel to
domain 9
[2005-07-08 10:17:07 xend] INFO (XendRoot:112) EVENT> xend.virq 4
[2005-07-08 10:17:07 xend] DEBUG (blkif:363) Unbinding vbd (type file) from
/dev/loop0
[2005-07-08 10:17:07 xend] INFO (XendRoot:112) EVENT>
xend.domain.exit['ttylinux2', '9', 'crash']
[2005-07-08 10:17:07 xend] INFO (XendRoot:112) EVENT>
xend.domain.destroy['ttylinux2', '9']
[2005-07-08 10:17:08 xend] INFO (XendRoot:112) EVENT>
xend.domain.died['ttylinux2', '9']

#############################################

The more interesting lines are these ones:

[2005-07-08 10:16:55 xend] DEBUG (XendDomainInfo:1130) Creating vbd dom=9
uname=file:/xen/vmttylinux2
[2005-07-08 10:16:55 xend] DEBUG (blkif:63) exception looking up device
number for sda1: [Errno 2] No such file or directory



I am using loop back devices to create my VMs, I have read somewhere that
this can cause troubles when it's used with NFS. And thats what I am doing,
but in "Live Migration Tutorial" by Michael McCabe, they use it without any
problems.


There are two machines in my network: XENmaster and XENslave. XENmaster is
running nfs-user-server, it exports a directory with (rw, sync,
no_root_squash) parameters. That directory has the loop-back-devices in it.
This is the VM's config file:

kernel = "/boot/vmlinuz-2.6.11-xenU"
memory = 8
name = "ttylinux2"
nics = 1
ip = "192.168.50.196 <http://192.168.50.196>"
disk = [ 'file:/xen/vmttylinux2,sda1,w' ]
root = "/dev/sda1 rw"

Ok, it's working*!!!!!!


































(*) not really. ;)

It works, but not totally. What I've done is to create a loopback device( in
the destiny machine) in the same path than the original one (in the origin
machine). Not using NFS now. It has the same size than the original one.
Now, I can migrate the domain. I see, after the migration, the domain in the
destiny machine and I can do "xm console" to that domain.
But something is wrong. That domain continues responding to ping but it
filesystem is corrupted!!!! I can't move in the directory tree.
Furthermore, when I look at the logs, I see the memory (ram) data moving
from one machine to another, but I can not see the data in the filesystem
doing that!!, and the time the migration lasts is too short to copy all the
filesystem through the network.

Should I use NFS? How do you people do this?

I promise this is my last message today.

Miguel Gómez wrote:

> Ok, it's working*!!!!!!
>
> (*) not really. ;)
>
> It works, but not totally. What I've done is to create a loopback device
> ( in the destiny machine) in the same path than the original one (in the
> origin machine). Not using NFS now. It has the same size than the
> original one. Now, I can migrate the domain. I see, after the migration,
> the domain in the destiny machine and I can do "xm console" to that
> domain.
>
> But something is wrong. That domain continues responding to ping but it
> filesystem is corrupted!!!! I can't move in the directory tree.
> Furthermore, when I look at the logs, I see the memory (ram) data moving
> from one machine to another, but I can not see the data in the
> filesystem doing that!!, and the time the migration lasts is too short
> to copy all the filesystem through the network.
>
> Should I use NFS? How do you people do this?
>
> I promise this is my last message today.

:-) Nice try.

Xen does not (currently) support synchronization of disk content. It relies
on you to do that. So you have to use NFS, or some other distributed system
like GFS or drbd to synchronize disks.

I am sorry I can't answer why the NFS setup does not work for you. I have
not yet tried myself.

Can you create the vm on both nodes (without migration) with the same config
file? Trying that would be the first thing to do for me.
Is the NFS file system mounted on both hosts in the same place?

Best Regards,
Michael Paesold


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