Closed bijenkins closed 5 months ago
xiehan
commented
11 months ago I'll leave it for the rest of the team to comment on the technical details, but FWIW this looks like a duplicate of #3111 (though this one has a lot more detail and should help the team with debugging; thank you!).
bijenkins
commented
11 months ago We are willing to work with whoever to get this resolved, we are using the CDKTF extensively and this is a blocker for us. Thank you @xiehan
bijenkins
commented
10 months ago @DanielMSchmidt could we schedule a time to meetup? I'm available whenever you are.
bijenkins
commented
8 months ago This has been sitting quite a while @xiehan can we get a status update by chance?
DanielMSchmidt
commented
8 months ago I tried to reproduce this and for me it worked fine with 0.20.1. For me it resulted in this cdk code (i made it a stack and copied it into a fresh project):
#!/usr/bin/env python
from constructs import Construct
from cdktf import App, TerraformStack
from cdktf import Token
#
# Provider bindings are generated by running `cdktf get`.
# See https://cdk.tf/provider-generation for more details.
#
from imports.nutanix.data_nutanix_cluster import DataNutanixCluster
from imports.nutanix.data_nutanix_clusters import DataNutanixClusters
from imports.nutanix.data_nutanix_image import DataNutanixImage
from imports.nutanix.data_nutanix_subnet import DataNutanixSubnet
from imports.nutanix.provider import NutanixProvider
from imports.nutanix.virtual_machine import VirtualMachine
class MyStack(TerraformStack):
def __init__(self, scope, name):
super().__init__(scope, name)
NutanixProvider(self, "nutanix",
endpoint="10.6.100.20",
insecure=True,
password="",
port=Token.as_string(9440),
username="terraform_admin"
)
cluster1 = DataNutanixCluster(self, "cluster1",
cluster_id="0005f342-2700-****-467a-*********"
)
DataNutanixClusters(self, "clusters")
s2019 = DataNutanixImage(self, "s2019",
image_id="e"
)
vlan618 = DataNutanixSubnet(self, "VLAN_618",
subnet_id="cd38c43c-fa7a-****-9bf0-***********"
)
VirtualMachine(self, "ggtest01",
cluster_uuid=Token.as_string(cluster1.cluster_id),
description="demo Frontend Web Server",
disk_list=[{
"data_source_reference": {
"kind": "image",
"uuid": Token.as_string(s2019.image_id)
},
"device_properties": {
"device_type": "DISK",
"disk_address": {
"adapter_type": "SCSI",
"device_index": Token.as_string(0)
}
},
"storage_config": {
"storage_container_reference": [{
"kind": "storage_container",
"uuid": "17a6c666-db20-4179-9a7c-*********"
}
]
}
}
],
memory_size_mib=16000,
name="test01",
nic_list=[{
"subnet_uuid": Token.as_string(vlan618.subnet_id)
}
],
num_sockets=1,
num_vcpus_per_socket=2
)
app = App()
MyStack(app, "tmp.axnj99KS6U")
app.synth()And I got this synthed JSON which to me looks also correct
{
"//": {
"metadata": {
"backend": "local",
"stackName": "tmp.axnj99KS6U",
"version": "0.20.1"
},
"outputs": {
}
},
"data": {
"nutanix_cluster": {
"cluster1": {
"//": {
"metadata": {
"path": "tmp.axnj99KS6U/cluster1",
"uniqueId": "cluster1"
}
},
"cluster_id": "0005f342-2700-****-467a-*********"
}
},
"nutanix_clusters": {
"clusters": {
"//": {
"metadata": {
"path": "tmp.axnj99KS6U/clusters",
"uniqueId": "clusters"
}
}
}
},
"nutanix_image": {
"s2019": {
"//": {
"metadata": {
"path": "tmp.axnj99KS6U/s2019",
"uniqueId": "s2019"
}
},
"image_id": "e"
}
},
"nutanix_subnet": {
"VLAN_618": {
"//": {
"metadata": {
"path": "tmp.axnj99KS6U/VLAN_618",
"uniqueId": "VLAN_618"
}
},
"subnet_id": "cd38c43c-fa7a-****-9bf0-***********"
}
}
},
"provider": {
"nutanix": [
{
"endpoint": "10.6.100.20",
"insecure": true,
"password": "",
"port": 9440,
"username": "terraform_admin"
}
]
},
"resource": {
"nutanix_virtual_machine": {
"ggtest01": {
"//": {
"metadata": {
"path": "tmp.axnj99KS6U/ggtest01",
"uniqueId": "ggtest01"
}
},
"cluster_uuid": "${data.nutanix_cluster.cluster1.cluster_id}",
"description": "demo Frontend Web Server",
"disk_list": [
{
}
],
"memory_size_mib": 16000,
"name": "test01",
"nic_list": [
{
}
],
"num_sockets": 1,
"num_vcpus_per_socket": 2
}
}
},
"terraform": {
"backend": {
"local": {
"path": "/private/var/folders/m4/673s3vwn1_g7c72bmvq521g00000gn/T/tmp.axnj99KS6U/terraform.tmp.axnj99KS6U.tfstate"
}
},
"required_providers": {
"nutanix": {
"source": "nutanix/nutanix",
"version": "1.9.5"
}
}
}
}Could you try again with the current cdktf version?
bijenkins
commented
7 months ago @DanielMSchmidt if you notice your synthesized json output has a empty disk_list and nic_list that is under resource.nutanix_virtual_machine.ggtest01. The issue is present in your example.
Those fields should not be empty, as you try to run that via a applyor deploy it will throw a error.
bijenkins
commented
6 months ago Looking for a status update as it's been months @DanielMSchmidt
github-actions[bot]
commented
4 months ago I'm going to lock this issue because it has been closed for 30 days. This helps our maintainers find and focus on the active issues. If you've found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.
Expected Behavior
TL;DR After converting the Nutanix Provider, following the provided docs, converting a vm resource from terraform using the cdktf convert tool errors on conversion, it's expected to convert the provider and convert the
main.tffilecat main.tf | cdktf convert --provider 'nutanix/nutanix' --language python > test.pyresults in a array error from theprovider-generator/ .../resource-parser.jsThis lines up with the behavior experienced if one converts the provider the themselves and constructs a Terraform Stack in Python code themselves. Using the Terraform stack in this way, results in array fields on the Nutanix VM array parameter fields being empty in the cdktf.out.json.
Read
The cdktf conversion process should be able to convert this Nutanix VM Terraform manifest. Following instructions from the cdktf convert subcommand.
The cdktf being used via Python, manually writing the code to create a Nutanix Terraform stack should convert all fields of a VM without blank fields.
These two different yet connected scenarios we believe incorrectly convert types of arrays for disk_list, and nic_list.
Actual Behavior
See steps to reproduce.
Steps to Reproduce
Manual Conversion of Provider and Running Created Stack code:
The manual conversion process is documented in a issue in the nutanix github space:
https://github.com/nutanix/terraform-provider-nutanix/issues/624
After contacting Nutanix they stated to go through Hashicorp Support due to them not supporting CDKTF. They are a trusted partner according to their registry page.
Automated Conversion Steps
Output
Note
I tested this on node 18 to node 20 using NVM.
Files:
cdktf.json
main.tf
Versions
cdktf debug language: python cdktf-cli: 0.18.0 node: v20.2.0 terraform: 1.5.6 arch: arm64 os: darwin 21.3.0
Providers
┌─────────────────┬──────────────────┬───────┬────────────┬──────────────┬─────────────────┐ │ Provider Name │ Provider Version │ CDKTF │ Constraint │ Package Name │ Package Version │ ├─────────────────┼──────────────────┼───────┼────────────┼──────────────┼─────────────────┤ │ nutanix/nutanix │ 1.9.3 │ │ │ │ │
Gist
No response
Possible Solutions
We believe the struct that nutanix has defined in their provider is incorrectly setting type, or that sequence is not being picked up by jsii.
Converted code using
cdktf get:Workarounds
There's none, we can't convert this module and use it using the provided conversion tools, and trying to create the python code manually results in unexpected blank
Sequencefields.Anything Else?
We also tested conversion using javascript instead of python with VERY similar errors.
References
No response
Help Wanted
Community Note