Existing deploy deployment type fails to complete.

[GlenWillms]

Describe the bug

When using the existing deploy template, the deployment fails to complete. All resources appear to be deployed however, the BIG-IP VEs do not complete provisioning. VE nodes are unable connect to the wireserver to complete the deployment. The VMs do not have the VM agent either.

Expected behavior

It is expected that provisioning will be sucessful and that the Azure VM agent would also be deployed.

Current behavior

VEs fail to provision and are in loop trying to connect to the wire server.

Possible solution

Unknown

Steps to reproduce

1. 1 VNet, with 3 subnets that match the deployment parameters file.
2. Deploy using the following option: https://portal.azure.com/#create/Microsoft.Template/uri/https%3A%2F%2Fraw.githubusercontent.com%2FF5Networks%2Ff5-azure-arm-templates-v2%2Fv2.9.0.0%2Fexamples%2Ffailover%2Fazuredeploy-existing-network.json
3. Deployment parameters file attached.

Screenshots

None

Context

Unable to complete VE provisioning. Project stalled.
armdeploy-parametersFile.json

Your Environment

Multiple Azure tenants. Can be reproduced in a brand new resource group.

[mikeshimkus]

@GlenWillms Can you connect to the VE instances using the provided ssh key and grab the contents of /var/log/cloud/startup-script.log and provide a sanitized version?

I see a couple of thing wrong right off the bat, in the params file no value was provided for either bigIpPasswordSecretId or bigIpPasswordSecretValue; you need to supply one of those two (see https://github.com/F5Networks/f5-azure-arm-templates-v2/tree/main/examples/failover#template-input-parameters). Also the instance type is Standard_D8s_v5, but the templates have only been tested up to v4.

[GlenWillms]

Thanks for the fast response.

The password was added at deployment time.

I will provide the log as soon as possible.

My full deploy worked using a v4 VM. I hadn't noticed that they were different before. I'll give that a shot next time.

[GlenWillms]

In speaking with the customer's consultant, a v4 image type was tried. v5 was attempted to try something else. I'll still try to repo with v4.