On 3/28/2018 4:35 PM, Jay Pipes wrote:
> On 03/28/2018 03:35 PM, Matt Riedemann wrote:
>> On 3/27/2018 10:37 AM, Jay Pipes wrote:
>>>
>>> If we want to actually fix the issue once and for all, we need to
>>> make availability zones a real thing that has a permanent identifier
>>> (UUID) and store that permanent identifier in the instance (not the
>>> instance metadata).
>>>
>>> Or we can continue to paper over major architectural weaknesses like
>>> this.
>>
>> Stepping back a second from the rest of this thread, what if we do the
>> hard fail bug fix thing, which could be backported to stable branches,
>> and then we have the option of completely re-doing this with aggregate
>> UUIDs as the key rather than the aggregate name? Because I think the
>> former could get done in Rocky, but the latter probably not.
>
> I'm fine with that (and was fine with it before, just stating that
> solving the problem long-term requires different thinking)
>
> Best,
> -jay
Just FYI for anyone that cared about this thread, we agreed at the Stein
PTG to resolve the immediate bug [1] by blocking AZ renames while the AZ
has instances in it. There won't be a microversion for that change and
we'll be able to backport it (with a release note I suppose).
[1] https://bugs.launchpad.net/nova/+bug/1782539
--
Thanks,
Matt
_______________________________________________
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
> On 03/28/2018 03:35 PM, Matt Riedemann wrote:
>> On 3/27/2018 10:37 AM, Jay Pipes wrote:
>>>
>>> If we want to actually fix the issue once and for all, we need to
>>> make availability zones a real thing that has a permanent identifier
>>> (UUID) and store that permanent identifier in the instance (not the
>>> instance metadata).
>>>
>>> Or we can continue to paper over major architectural weaknesses like
>>> this.
>>
>> Stepping back a second from the rest of this thread, what if we do the
>> hard fail bug fix thing, which could be backported to stable branches,
>> and then we have the option of completely re-doing this with aggregate
>> UUIDs as the key rather than the aggregate name? Because I think the
>> former could get done in Rocky, but the latter probably not.
>
> I'm fine with that (and was fine with it before, just stating that
> solving the problem long-term requires different thinking)
>
> Best,
> -jay
Just FYI for anyone that cared about this thread, we agreed at the Stein
PTG to resolve the immediate bug [1] by blocking AZ renames while the AZ
has instances in it. There won't be a microversion for that change and
we'll be able to backport it (with a release note I suppose).
[1] https://bugs.launchpad.net/nova/+bug/1782539
--
Thanks,
Matt
_______________________________________________
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators