[one-users] WG: Problem with (live)migration

Thu Mar 11 06:22:12 PST 2010

Hi,

the problem why I couldn't add the machines as nodes properly was the
language configuration of them. The locales were configured for de_DE.utf8
so OpenNebula couldn't get the information right from the nodes since "virsh
nodeinfo" executed on the nodes has an output in the system language.

Now I'm facing new issues. When livemigrating a VM everything works fine but
afterwards the VM isn't reachable/pingable anymore. Even after
relivemigrating back to the original node. All nodes have the network
bridges configured and are monitored successfully by the oned.

We tried VNC access but there you have to configure the IP of the host in
the VM template. But when migrating this IP changes. Is there a possibility
to have access via VNC after migrating the VM from one node to another?

Best,

Stefan

> -----Ursprüngliche Nachricht-----
> Von: Joseph Glanville [mailto:shortstuff92 at gmail.com]
> Gesendet: Dienstag, 9. März 2010 19:28
> An: Harder, Stefan
> Cc: Jack Jill; users at lists.opennebula.org
> Betreff: Re: [one-users] WG: Problem with (live)migration
> 
> Hi,
> 
> Ensure you have ruby properly installed on the machine `brandy` and
> have configured ssh public key authentication without a passphrase for
> the oneadmin account.
> That should help you retrieve stats from `brandy` and hopefully
> resolve your migration issues.
> If you try deploy an instance to `brandy` is there anything written to
> /srv/cloud/var/<vmid>/*.log ?
> Also check /srv/cloud/var/oned.log for "Host Monitoring" errors if the
> above doesn't allow you to retrieve stats from `brandy`.
> 
> Hope that helps,
> 
> Kind regards,
> 
> Joseph Glanville
> 
> GPG Key ID: A5FC8BE3
> 
> 
> 
> On Tue, Mar 9, 2010 at 1:28 AM, Harder, Stefan
> <Stefan.Harder at fokus.fraunhofer.de> wrote:
> > Hi,
> >
> >
> >
> > the only output I get from the log when I try to add “brandy” as node
> is:
> >
> >
> >
> > Mon Mar  8 15:02:24 2010 [ReM][D]: HostAllocate method invoked
> >
> >
> >
> > Deploying VMs on “vodka” works fine.
> >
> >
> >
> > Best,
> >
> >
> >
> > Stefan
> >
> >
> >
> >
> >
> > Von: Jack Jill [mailto:jacknjill111 at gmail.com]
> > Gesendet: Donnerstag, 25. Februar 2010 21:34
> > An: Harder, Stefan
> > Cc: users at lists.opennebula.org
> > Betreff: Re: [one-users] WG: Problem with (live)migration
> >
> >
> >
> >
> >
> > On Thu, Feb 25, 2010 at 9:01 AM, Harder, Stefan
> > <Stefan.Harder at fokus.fraunhofer.de> wrote:
> >
> > I have some additions to my last Email:
> >
> > I have two machines "brandy" on which the cloud manager is installed
> and
> > "vodka" the cluster node. I created a vnet and then the ttylinux
> example and
> > it worked after some try and error.
> >
> > For testing the migration functionality I wanted to add the brandy
> machine
> > also as cluster node with
> >
> > oneadmin at brandy:/srv/cloud$ onehost create brandy im_kvm vmm_kvm
> tm_nfs
> >
> > After a while onehost list shows:
> >
> >  ID NAME                      RVM   TCPU   FCPU   ACPU    TMEM
>  FMEM STAT
> >   6 vodka                       1    400    398    398 8189944
> 7667316   on
> >   7 brandy                      0    400    400    400       0
> 7859184   on
> >
> > Without the TMEM info. And migration to this host doesn’t work as
> mentioned
> > in my first email. If I delete the vodka machine from the hosts and
> want to
> > deploy a vm on brandy it doesn’t work, the vm stays in pend status.
> If I
> > deploy a vm on brandy without OpenNebula only with virsh it works. So
> I
> > think something went wrong creating the host brandy. But what?
> >
> > Could you please post relevant (host brandy creation) log output from
> > oned.log?
> >
> > Can you deploy VMs on host vodka?
> >
> >
> > Any guesses?
> >
> > Best,
> >
> > Stefan
> >
> >
> >> -----Ursprüngliche Nachricht-----
> >> Von: Harder, Stefan
> >> Gesendet: Donnerstag, 25. Februar 2010 10:40
> >> An: 'users at lists.opennebula.org'
> >> Betreff: Problem with (live)migration
> >
> >>
> >> Hi,
> >>
> >> I get the following error when I try to migrate or livemigrate a
> >> running VM (in this case the ttylinux example) from one host to
> >> another. They have a shared FS which is under /srv/cloud.
> >>
> >> Thu Feb 25 10:34:56 2010 [DiM][D]: Live-migrating VM 50
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: LOG - 50
> Command
> >> execution fail: virsh --connect qemu:///system migrate --live one-50
> >> qemu+ssh://brandy/session
> >>
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: LOG - 50 STDERR
> >> follows.
> >>
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: LOG - 50
> >> /usr/lib/ruby/1.8/open3.rb:67: warning: Insecure world writable dir
> >> /srv/cloud/one in PATH, mode 040777
> >>
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: LOG - 50
> >> Connecting to uri: qemu:///system
> >>
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: LOG - 50 error:
> >> operation failed: domain with the same name or UUID already exists
> as
> >> 'one-50'
> >>
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: LOG - 50
> ExitCode:
> >> 1
> >>
> >> Thu Feb 25 10:34:57 2010 [VMM][D]: Message received: MIGRATE FAILURE
> 50
> >> -
> >>
> >>
> >> Does somebody know why the domain already eyists?
> >>
> >> Best Regards,
> >>
> >> Stefan
> >
> > _______________________________________________
> > Users mailing list
> > Users at lists.opennebula.org
> > http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
> >
> >
> > --
> > Thanks,
> > Rosh
> > http://blogs.plexibus.com
> >
> > _______________________________________________
> > Users mailing list
> > Users at lists.opennebula.org
> > http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
> >
> >
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 7099 bytes
Desc: not available
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20100311/897825a6/attachment-0003.bin>