Occassional "Instance failed to spawn" in contrail 2.21
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Juniper Openstack |
New
|
High
|
Rudra Rugge |
Bug Description
COpying the email exchange to this bug :
Using contrail 2.21 on Ubuntu 14.04
===
From: Praneet Bachheti
Sent: Thursday, October 15, 2015 2:26 PM
To: Ashish Ranjan <email address hidden>
Cc: PrasannaKumar Anandan Gajendran <email address hidden>; Deepak Jayaraman <email address hidden>; Ranjini Rajendran <email address hidden>; Sreelakshmi Sarva <email address hidden>; Arivudainambi A <email address hidden>; Sudhir Kumar <email address hidden>; Hari Prasad Killi <email address hidden>; Rudra Rugge <email address hidden>
Subject: Re: Heat stack-create failure in Contrail 2.21
I think we hit a bug … here is the explanation and a workaround for now until it is fixed …
The VM created by the service-instance is not present in Nova although the VM id is present in the api-server.
Whenever we do a stack-show, heat tried to get the VM from nova and fails throwing an exception of NOT_FOUND
Talked to Rudra, it is possible sometime that request to nova for creating VM may not succeed. In that case, we will never relaunch the VM and stale vm-id will be left in api-server.
Workaround:
delete the virtual-machine from api-server, and the VM will be relaunched by service-monitor.
command to do that
curl -u admin:contrail123 -X delete http://
The id is the end is the instacne-id which is returned by heat error log.
For now I have applied the workaround and I can do a heat stack-show.
- Praneet
On Oct 15, 2015, at 1:41 PM, Praneet Bachheti <email address hidden> wrote:
Looking in to it.
On Oct 15, 2015, at 1:28 PM, Ashish Ranjan <email address hidden> wrote:
Praneet could you check.
On Oct 15, 2015, at 1:24 PM, PrasannaKumar Anandan Gajendran <email address hidden> wrote:
Ashish,
Can someone please take a look at this ASAP.
Thanks
From: Deepak Jayaraman
Sent: Thursday, October 15, 2015 1:22 PM
To: Ranjini Rajendran <email address hidden>; Sreelakshmi Sarva <email address hidden>; Praneet Bachheti <email address hidden>
Cc: PrasannaKumar Anandan Gajendran <email address hidden>; Arivudainambi A <email address hidden>; Sudhir Kumar <email address hidden>; Hari Prasad Killi <email address hidden>
Ranjini , Sree,
We re-provisioned the server nethra-01 with 2.21 again and we ran into the same issue again.
The setup nethra-01 (root/Embe1mpls) is in this state, please debug so that we can quickly identify the root cause.
Test:
We were bringing up 20 heat-stacks serially with 10sec interval - 16th one failed and is in this state:
the 2 instances after the failed one are fine ...
| id | stack_name | stack_status | creation_time |
| fbb9c12f-
| 492c13d4-
| 526f6271-
| 2b8fb47f-
| 7011d204-
| 149f1d40-
| 544eca81-
| a1dee4ac-
| 18313a51-
| 09e297bc-
| 916c387a-
| 788dae21-
| 26afd46f-
| 6f6f99a2-
| a92c8981-
| ac9d8816-
| d14ad529-
| 655dae6b-
ERROR: Remote error: NotFound Instance could not be found (HTTP 404) (Request-ID: req-9b4a1f23-
But the other ones are fine :
| description | HOT template to create network service |
| | "output_value": "708d8fa9-
| | "description": "Service template name", |
| | "output_key": "vFF-0-
| | "description": "Service template service instances", |
| | "output_key": "vFF-0-
| | "output_value": "default-
| | "description": "Service instance service template", |
| | "output_key": "vFF-0-
| | "output_value": "default-
| | "description": "Service template FQ name", |
| | "output_key": "vFF-0-
| template_
Diving into the logs:
nova log:
Changed in juniperopenstack: | |
importance: | Undecided → High |
assignee: | nobody → Praneet Bachheti (praneetb) |
tags: | added: config heat |
Changed in juniperopenstack: | |
assignee: | Praneet Bachheti (praneetb) → Rudra Rugge (rudrarugge) |
information type: | Proprietary → Public |