- Add documentation for resources
- Rename files to match standard patterns
- Add acceptance tests for resource groups
- Add acceptance tests for vnets
- Remove ARM_CREDENTIALS file - as discussed this does not appear to be
an Azure standard, and there is scope for confusion with the
azureProfile.json file which the CLI generates. If a standard emerges
we can reconsider this.
- Validate credentials in the schema
- Remove storage testing artefacts
- Use ARM IDs as Terraform IDs
- Use autorest hooks for logging
This commit cleans up some of the work on the Azure ARM provider
following review by @phinze. Specifically:
- Unnecessary ASM-targeted tests are removed
- Validation is added to the `resource_group` resource
- `dns_servers_names` -> `dns_servers` as per the API documentation
- AZURE_SUBSCRIPTION_ID environment variable is renamed to be
ARM_SUBSCRIPTION_ID in order to match the other environment variables
This commit brings some of the work over from #3808, but rearchitects to
use a separate provider for Azure Resource Manager. This is in line with
the decisions made by the Azure Powershell Cmdlets, and is important for
usability since the sets of required fields change between the ASM and
ARM APIs.
Currently `azurerm_resource_group` and `azurerm_virtual_network` are
implemented, more resources will follow.
Each call to the Kinesis DescribeStream API returns a limited number of
shards. When interrogating AWS for the state of a Kinesis stream, the
client needs to page through the API's responses to get the true number
of shards.
It seems that not all Rundeck servers consistently return the project name
when retrieving a job. Not yet sure in what situations it is or isn't
returned, but since jobs are not allowed to move between projects anyway
it doesn't hurt to just skip refreshing it if the server provides no
value.
vmImageClient.ListVirtualMachineImages takes a parameter as of
68d50cb53a73edfeb7f17f5e86cdc8eb359a9528 in Azure/azure-sdk-for-go .
Passing in a parameters object whose members are all empty strings seems
to restore the previous behavior.
This change better reflects how block devices are passed to the Nova API
and allows for future enablement of block_device features. It also resolves
an interpolation bug.
James Nugent