Error with Invoking HTTP with Azure

[mousman85]

Hi,

I'm getting an intermitment error with Invoking a HTTP request to Azure. I've created a group and using the invoke HTTP to create a team for that group, but randomly getting the below error:

{
  "error": {
    "code": "NotFound",
    "message": "Failed to execute MS Graph backend request GetGroupInternalApiRequest. The server failed to respond correctly. Response Code: NotFound, Reason: Reason: Not Found, Correlation headers: request-id: 86438d80-68a6-4583-86c6-62fe44989777\r\nclient-request-id: 86438d80-68a6-4583-86c6-62fe44989777\r\nx-ms-ags-diagnostic: {\"ServerInfo\":{\"DataCenter\":\"UK South\",\"Slice\":\"SliceC\",\"Ring\":\"4\",\"ScaleUnit\":\"001\",\"RoleInstance\":\"AGSFE_IN_21\"}}\r\nDate: Fri, 21 Aug 2020 13:01:40 GMT\r\n",
    "innerError": {
      "date": "2020-08-21T13:01:40",
      "request-id": "ce6dd3ca-2d03-4d26-b124-6d557df8eb93"
    }
  }
}

This is isn't on every request just sometimes. Any suggestions on what this might be? 

The command for the invoke looks like this:

[rimatos]

HI @mousman85 ,

Taking a look at the Graph API documentation it could be related with the permissions for this request:

How are you authenticating against Graph in Azure? Do you have the application registered in Azure?

I ask this because I do not see any usage of the Authorization Header for authentication, which is needed when using the regular HTTP action. Were the credentials defined when creating the connection to the connector?

Regards,

Ricardo

[mousman85]

Hi, Its using a premium connector, i have a premium account, the auth is done on the login on the connector: 

I have tried a new connection as well and removing and re-adding this connector. Like i said it isn't all the time some creations go through fine create the team etc some don't, and i can't see see a pattern in the ones that fail. 

I have now added in a delay of 10-20 seconds before trying to create the team, in-case the group creation took longer and its not able to find the group before it adds the team.

[rimatos]

Given that the format of the request appears to be created properly and that the issue is only intermittent, I would recommend raising a ticket to Microsoft Support through https://admin.microsoft.com/AdminPortal/Home#/support/requests to our Graph team, that handles the API requests, so this behavior can be analyzed in more detail.

I just wanted to ask you 2 questions first:

1. Do you find this intermittent issue to occur whenever the group has already been created for a while?
2. When using a separate client (e.g. POSTMAN) did you ever faced any similar issues to this?

Regards,

Ricardo

[mousman85]

Thanks for the reply,

The error is only on new group creation, it all happens at one point, group is created members are added i've then added in a delay to then create the team. I haven't tried it with an existing group that has no team as we don't process those in the flow. 

It really is intermittent it failed again this morning but worked twice before and i can't see a pattern in the failing. I had another flow running a similar process to create a new team on a request but this was the same graph call but created the team at the same time as time as the group and i had the same error appear here as well, and this was using a post request in the invoke :

This one below and i'm not sure whats different creates a group and a team at the same time I do't' need to run a PUT request for a team. 

[rimatos · Accepted solution]

Hi @mousman85 ,

Looking into the Graph API documentation I discovered the following: 

You mentioned you introduced a 10/20 second delay, but did you include the retries as well?
It states in the documentation that this can occur if the group was created 15 min before creating the Team, so it appears to be the issue you are facing.

[mousman85]

Ah ha!

Thank you, that would explain why its intermittent, is there way to retry a section that fails, or would it be just create the invoke three times and change the on run to when failed? 

[rimatos]

I believe this could be done in the retry policy that is found in the Action card settings.

If that does not work you can try with the run to when failed.