Open Stack

Sorry, I misread the commit out of excitement for a fix. The one I 
mentioned is already applied in queens. But the exceptions we see differ 
from the one catched in https://review.opendev.org/#/c/675496/

Creating a node in queens is fine and doesn't raise the DBDuplicateEntry 
exception, because the UUID of the ironic node is actually the 
hypervisor_hostname of the compute node in queens - might have changed 
later on. So a unique index on the compute node's UUID doesn't help.

Johannes

On 6/22/20 10:09 AM, Johannes Kulik wrote:
> Hi Fred,
> 
> we're also on queens and face the same problem.
> 
> Looking at the linked patch from Mohammed, the code relies on 
> https://github.com/openstack/nova/commit/6865baccd3825bf891a763627693b8b299e524df 
> to work. So I guess, we will apply both in our environment.
> 
> Have a nice day,
> Johannes
> 
> On 6/20/20 2:22 AM, fsbiz at yahoo.com wrote:
>>   Thanks Mohammed.Seems relevant but the patches were applied only to 
>> train, ussuri and master.
>>
>> I'm still on Queens.  Not sure if the patches are relevant to Queens 
>> also.
>> Regards,Fred.
>>      On Friday, June 19, 2020, 01:59:27 PM PDT, Mohammed Naser 
>> <mnaser at vexxhost.com> wrote:
>>   Hi Fred:
>> You'll need to update to grab this 
>> change: https://review.opendev.org/#/c/675496/
>> Thanks,Mohammed
>>
>>
>>
>> On Thu, Jun 18, 2020 at 2:29 PM fsbiz at yahoo.com <fsbiz at yahoo.com> wrote:
>>
>> Hi folks,
>> Every time I get an ironic switchover I end up with a few resource 
>> provider errors as follows.
>>
>> 2020-06-10 05:11:28.129 75837 INFO nova.compute.resource_tracker 
>> [req-eac491b2-dd72-4466-b37e-878dbf40cda5 - - - - -] Compute node 
>> record created for 
>> sc-ironic06.nvc.nvidia.com:7c95f255-3c54-46ab-87cf-0b1707971e9c with 
>> uuid: 1b0b74d2-d0d4-4637-b7ea-3adaa58cac2e
>> 2020-06-10 05:11:28.502 75837 ERROR nova.scheduler.client.report 
>> [req-eac491b2-dd72-4466-b37e-878dbf40cda5 - - - - -] 
>> [req-d8af2589-8c75-427c-a7b1-d5270840a4c8] Failed to create resource 
>> provider record in placement API for UUID 
>> 1b0b74d2-d0d4-4637-b7ea-3adaa58cac2e. Got 409: {"errors": [{"status": 
>> 409, "request_id": "req-d8af2589-8c75-427c-a7b1-d5270840a4c8", 
>> "detail": "There was a conflict when trying to complete your 
>> request.\n\n Conflicting resource provider name: 
>> 7c95f255-3c54-46ab-87cf-0b1707971e9c already exists.  ", "title": 
>> "Conflict"}]}.
>> 2020-06-10 05:12:49.463 75837 ERROR nova.scheduler.client.report 
>> [req-eac491b2-dd72-4466-b37e-878dbf40cda5 - - - - -] 
>> [req-ffd13abc-08f3-47cd-a224-a07183b066ec] Failed to create resource 
>> provider record in placement API for UUID 
>> 1b0b74d2-d0d4-4637-b7ea-3adaa58cac2e. Got 409: {"errors": [{"status": 
>> 409, "request_id": "req-ffd13abc-08f3-47cd-a224-a07183b066ec", 
>> "detail": "There was a conflict when trying to complete your 
>> request.\n\n Conflicting resource provider name: 
>> 7c95f255-3c54-46ab-87cf-0b1707971e9c already exists.  ", "title": 
>> "Conflict"}]}.
>>
>> So far the only way that works for me to fix these is to un-enroll and 
>> then re-enroll the node.
>> Is there a simpler way to fix this?
>> Thanks,Fred.
>>
>>
>>
> 

-- 
Johannes Kulik
IT Architecture Senior Specialist
*SAP SE *| Rosenthaler Str. 30 | 10178 Berlin | Germany