Cannot delete infrastructure: 'Unsupported Auth Version requested

[BorjaEst]

Current behaviour

The im-web allows the creation of infrastructures with unsuported Auth Version but does not allow to destroy them.

Using the following template simple-node (with line 81 uncommented) .

Output:

aefd1abc-74b7-11ea-8dbb-0242ac110002 | 0 | Show | Show | failed | N/A |   |  
-- | -- | -- | -- | -- | -- | -- | --

If I press delete:

Error: Error: Code: 400. Error Destroying Inf: Error destroying the infrastructure:

<LibcloudError in None 'Unsupported Auth Version requested'>

Desired behaviour

Not sure if it is an issue on the destroying process, but I should be able to delete any infrastructure in my "Infrastructure" menu.

[micafer]

hi @BorjaEst,

Could you send what are the credentials that have you set in the im-web portal?

[BorjaEst]

Sure:

• host = https://ostserver:5000; => https://fedcloud-osservices.egi.cesga.es/:5000/v3
• username = egi.eu; => egi.eu
• tenant = openid; => openid
• password = egi_aai_token_value; => <My token from EGI got at https://aai.egi.eu/oidc/manage/user/services >
• auth_version = 3.x_oidc_access_token; => 3.x_oidc_access_token
• domain = project_name; => eosc-synergy.eu

This is the log file:

:: Contextualization Log ::

2020-04-01 12:33:40.276600: All VMs failed. No contextualize.
2020-04-01 12:33:40.280091: Error launching the VMs of type simple_node to cloud ID CESGA of type OpenStack. Error, no concrete system to deploy: simple_node in cloud: CESGA. Check if a correct image is being used

VM 0:
Error launching the VMs of type simple_node to cloud ID CESGA of type OpenStack. Error, no concrete system to deploy: simple_node in cloud: CESGA. Check if a correct image is being used

During the launch phase
The error was probably that the image was not correct and in the host definition here;

• host = https://ostserver:5000; => https://fedcloud-osservices.egi.cesga.es/:5000**/v3**
  However it is an inconvenient when I have a mistake that I cannot delete the infrastructure. Now I am having 10 infrestructures with status error whcih I cannot delete (even after changing the host).

[micafer]

The problem seems to be related with the image.
Adding /v3 to the url of the host is not a problem as IM ignores it.
In the creation you do not get the libcloud error as the IM is not even trying to create any VM.
In the deletion process it tries to connect to the provider to assure that there is not resource to delete and in this point it is failing with the libcloud error.

What IM service version are you using, and if it is possible what libcloud version.

[BorjaEst]

I am using the container (1.5.13, latest (Dockerfile))
https://github.com/grycap/im-web/blob/v1.5.13/docker/Dockerfile

im-web:
    image: grycap/im-web

For the libcloud, seem it is not installed:

$ docker exec im-web python -c "import libcloud ; print libcloud.__version__"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ImportError: No module named libcloud
$ docker exec im-web libcloud -v
OCI runtime exec failed: exec failed: container_linux.go:349: starting container process caused "exec: \"libcloud\": executable file not found in $PATH": unknown

[micafer]

I refer to the IM service, not the IM web portal.

[BorjaEst]

Sorry:

  im:
    image: grycap/im:1.9.1

$ docker exec im libcloud -v
OCI runtime exec failed: exec failed: container_linux.go:349: starting container process caused "exec: \"libcloud\": executable file not found in $PATH": unknown

$ docker exec im python -c "import libcloud ; print libcloud.__version__"
OCI runtime exec failed: exec failed: container_linux.go:349: starting container process caused "exec: \"python\": executable file not found in $PATH": unknown

$ docker exec im python3 -c "import libcloud ; print libcloud.__version__"
  File "<string>", line 1
    import libcloud ; print libcloud.__version__
                                   ^
SyntaxError: invalid syntax

[micafer]

HI @BorjaEst,

Could try to restart the IM service (not the IM web portal (and try to delete the infras).

If you set a correct image, for example this one:

'appdb://CESGA/egi.docker.ubuntu.16.04s?eosc-synergy.eu'

The VM creation works?

[BorjaEst]

Hi @micafer,
Restart the IM service does not help (my data is persistent in a local folder).
However, if the image is correct, the instance is correctly deployed 👍 (I tried using ost:/server).

fce527f2-77f4-11ea-8f67-0242ac110003	0	Show	Show	unconfigured

The "unconfigured" is because I still need to fix something else:

2020-04-06 10:58:08.871570: Select master VM
2020-04-06 10:58:08.872569: Contextualization Error: No correct Master VM found. Check if there a linux VM with Public IP.
2020-04-06 10:58:08.922163: Error in the ansible installation process: No master VM found.
2020-04-06 10:58:13.925544: Error in the ansible installation process: No master VM found.
2020-04-06 10:58:33.931975: Error in the ansible installation process: No master VM found.

[micafer]

Hi @BorjaEst ,

I assume that your data is persistent, but restarting the IM clean some in memory data that I wanted to assure that is not the origin of your problem.
So now that is restarted:

• Have you tried to delete the incorrect infrastructures?
• Did you get the same error?

Regarding the unconfigured state.
Could you send me the information about the launched VM provided by the IM portal?

[BorjaEst]

Hi @micafer,
Yes, I tried, but I get the same error:

Error: Error: Code: 400. Error Destroying Inf: Error destroying the infrastructure:
No auth data has been specified to OpenStack.

Regarding the unconfigured state:
The provided outputs:

Outputs
| node_ipnode_credsuser = cloudadmtoken_type = private_keytoken = -----BEGIN RSA PRIVATE KEY----- ... -----END RSA PRIVATE KEY----- 
| node_ip |   
| node_creds 
| user = cloudadmtoken_type = private_keytoken = -----BEGIN RSA PRIVATE KEY----- ... -----END RSA PRIVATE KEY-----

-- | -- | -- | -- | -- | --
node_ip |  
node_creds 
| user = cloudadmtoken_type = private_keytoken = -----BEGIN RSA PRIVATE KEY----- ... -----END RSA PRIVATE KEY-----

The tosca template:

tosca_definitions_version: tosca_simple_yaml_1_0

imports:
  - indigo_custom_types: https://raw.githubusercontent.com/indigo-dc/tosca-types/master/custom_types.yaml

description: >
  Launch a compute node getting the IP and SSH credentials to access via ssh
metadata:
  display_name: Start a virtual machine
  icon: https://indigo-paas.cloud.ba.infn.it/public/images/vm-icon.png

topology_template:

  node_templates:

    simple_node:
      type: tosca.nodes.indigo.Compute
      capabilities:
        endpoint:
          properties:
            network_name: net-eosc-synergy.eu
        scalable:
          properties:
            count: 1
        host:
          properties:
            instance_type: cor1mem1hd10
        os:
          properties:
            image: ost://fedcloud-osservices.egi.cesga.es:9292/0fb385b2-f4a5-4a4f-8058-f21d4e78b7fa

  outputs:
    node_ip:
      value: { get_attribute: [ simple_node, public_address, 0 ] }
    node_creds:
      value: { get_attribute: [ simple_node, endpoint, credential, 0 ] }

[micafer]

Hi @BorjaEst ,

the error

Error: Error: Code: 400. Error Destroying Inf: Error destroying the infrastructure:
No auth data has been specified to OpenStack.

Means that you have not provided any OpenStack auth to the site where the infras was deployed.

And can you provide me the VM info? Click on the "0" ID that should appear in the second column (titled "VM IDs").

[BorjaEst]

Hi, sorry but I tried to reproduce the error again but I was not able.