Open Stack

> Just a thought: I think there should be some sort of scheduling intelligence
> involved with a command like this. Right now, all of the migration and
> evacuation commands work on individual instances being moved to a specified
> compute node.

> Or does this intelligence already exist in OpenStack and I'm just falling
> behind?

The Host parameter is optional, if you omit it, the scheduler chooses a host.

# nova live-migration
usage: nova live-migration [--block-migrate] [--disk-over-commit]
                           <server> [<host>]


(accidentally sent from the wrong email the first time I sent this and
the list rejected it, apologies if you get multiple copies)

On Fri, Feb 14, 2014 at 8:49 AM, Joe Topjian <joe at topjian.net> wrote:
> I agree.
>
> Just a thought: I think there should be some sort of scheduling intelligence
> involved with a command like this. Right now, all of the migration and
> evacuation commands work on individual instances being moved to a specified
> compute node. IMO, automating this should involve the evacuation going to
> the scheduler to figure out what other compute nodes are appropriate as
> destinations.
>
> Without this feature, if you were to do something like:
>
> for i in $instance_list
> do
>   nova live-migration $i c01
> end
>
> c01 might already be at capacity or would be well-over capacity by the time
> the loop finishes.
>
> Individual sites might make their own variation of this (such as choosing a
> compute node based on the nth instance being migrated) but that might not be
> the most "professional" solution :)
>
> Or does this intelligence already exist in OpenStack and I'm just falling
> behind?
>
>
>
> On Fri, Feb 14, 2014 at 1:47 PM, Tim Bell <Tim.Bell at cern.ch> wrote:
>>
>> At minimum, getting something into stackforge would be good.
>>
>>
>>
>> Recovering as best you can from a dead hypervisor situation is another
>> process that we need to define best-of-breed scripts.
>>
>>
>>
>> Tim
>>
>>
>>
>>
>>
>> From: Robert van Leeuwen [mailto:Robert.vanLeeuwen at spilgames.com]
>> Sent: 14 February 2014 12:50
>> To: Tim Bell
>> Cc: openstack-operators at lists.openstack.org
>> Subject: RE: [Openstack-operators] Putting a hypervisor into maintenance /
>> how to get a list of all instance IDs on a host
>>
>>
>>
>> > How about sharing the script if you can get it to a generic state ? This
>> > would seem to be an activity that many
>> > operators would like to be able to do.
>> > Tim
>>
>>
>>
>> I *think* it should become part of the Openstack tools just like the
>> evacuate command.
>>
>> I'm currently hacking something together in bash for a quick fix.
>> If it will ever get into a share-able state I'll make sure to do so :)
>>
>> Hopefully I will have some time to create a proper python version.
>> When that happens I'll try to get it into the nova project.
>>
>> Cheers,
>> Robert van Leeuwen
>>
>>
>> _______________________________________________
>> OpenStack-operators mailing list
>> OpenStack-operators at lists.openstack.org
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>
>
>
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>