[one-users] OpenNebuka Hosted-VLAN support
Patrice LACHANCE
patlachance at gmail.com
Mon Oct 17 08:59:46 PDT 2011
Hello
In previous version there was a cluster feature that was replaced in
OpenNebula 3.0 by ozones.
Shouldn't opennebula make sure that all the nodes in a zone are able to run
a vm and thus handle network creation on all nodes before starting a new VM?
Patrice
2011/9/27 Alberto Picón Couselo <alpicon1 at gmail.com>
> **
> Hello,
>
> We are testing hosted VLAN support in OpenNebula to implement network
> isolation. This feature seems to work correctly when a new instance is
> deployed, as it is stated in oned.conf, hm-vlan hook is executed in PROLOG
> state.
>
> However, there are another states where VLANs and bridges should be created
> (or its existence checked) before executing a concrete operation:
>
> * Migration/Live migration of an instance to a hypervisor where VLAN and
> bridge of the instance has never been created
> VLAN and bridge existence should be checked and created if necessary before
> migration is executed. Opennebula 3.0 RC1 performs migration without doing
> these checks and fails to migrate/live migrate the instance, leaving it in a
> FAILED state.
>
> * A failed instance cannot be redeployed to a hypervisor where VLAN and
> bridge of the instance has never been created
> VLAN and bridge existence should be checked and created if necessary to
> redeploy the image to the selected hypervisor.
>
> * A stopped instance cannot be resumed if VLAN and bridge of the instance
> does not exist.
> If we stop all instances of a concrete hypervisor and reboot the hypervisor
> for maintenance purposes, all bridges and VLANs will be deleted. Stopped
> instances won't resume because VLANs and bridges requirements are not
> satisfied and will enter in a FAILED state (performing a deletion of non
> persistent disks; BTW, we have removed deletion lines in tm_delete script
> for the moment, :D).
>
> So, VLAN and bridge existence should be checked and created if necessary
> to resume/migrate/livemigrate/recover_from_failed_state the instance to the
> selected hypervisor. As it is stated in oned.conf, hm-vlan hook could be
> executed on:
>
> # Virtual Machine Hooks (VM_HOOK) defined by:
> # name : for the hook, useful to track the hook (OPTIONAL)
> # on : when the hook should be executed,
> # - CREATE, when the VM is created (onevm create)
> # - PROLOG, when the VM is in the prolog state
> # - RUNNING, after the VM is successfully booted
> # - SHUTDOWN, after the VM is shutdown
> # - STOP, after the VM is stopped (including VM image
> transfers)
> # - DONE, after the VM is deleted or shutdown
> # - FAILED, when the VM enters the failed state
>
> But I'm not able to find a procedure to implement these functionalities in
> oned.conf for the states I mentioned.
>
> Please, can you give me any clues?
>
> Best Regards,
> Alberto Picón
>
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opennebula.org
> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>
>
--
Patrice LACHANCE
Manager IT Consulting, Logica : http://www.logica.com
Réseau Viaduc
Consultez mon profil:
http://www.viaduc.com/public/profil/?memberId=00226pj42r07h9f3
Vous inscrire sur le réseau:
http://www.viaduc.com/invitation/00226pj42r07h9f3
LinkedIn Network:
See my profile: http://www.linkedin.com/in/plachance
Join the network: http://www.linkedin.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20111017/6970b747/attachment.htm>
More information about the Users
mailing list