Interactive Terraform visualization. State and configuration explorer.

Posting this for now as I continue to debug, but maybe others are seeing similar issues.

Overview: Runtime error when attempting to generate graph from provided plan that points to an invalid memory address reference. Upon further debugging, it appears to be related to a Data source type being labeled as a Resource source type. At the moment I am unsure of how to reproduce.

Release Version: 3.0

ERROR Message:

> rover -planPath plan.out
2022/01/31 09:48:08 Starting Rover...
2022/01/31 09:48:08 Using provided plan...
2022/01/31 09:48:12 Generating resource overview...
2022/01/31 09:48:12 Generating resource map...
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x48 pc=0x174c28e]

goroutine 1 [running]:
main.(*rover).GenerateModuleMap(0xc000187680, 0xc00009fdd0, 0xc00036ac40, 0x38)
	/Users/joseestrada/rover/map.go:225 +0x1dae
main.(*rover).GenerateModuleMap(0xc000187680, 0xc00009e5a0, 0xc0003a8cf0, 0x25)
	/Users/joseestrada/rover/map.go:253 +0x2357
main.(*rover).GenerateModuleMap(0xc000187680, 0xc00050e750, 0xc000399170, 0x12)
	/Users/joseestrada/rover/map.go:253 +0x2357
main.(*rover).GenerateModuleMap(0xc000187680, 0xc000535b88, 0x0, 0x0)
	/Users/joseestrada/rover/map.go:253 +0x2357
main.(*rover).GenerateMap(0xc000187680, 0x0, 0x0)
	/Users/joseestrada/rover/map.go:335 +0x1ee
main.(*rover).generateAssets(0xc000187680, 0xc000187680, 0x2)
	/Users/joseestrada/rover/main.go:193 +0x130
main.main()
	/Users/joseestrada/rover/main.go:138 +0xea5

Additional Context: From what I gather and understand, it appears that my generated Terraform Plan is causing a runtime error due to a resource being mislabeled. I believe that a Data source is being labeled as a Resource type and thus causing the error.

This error is happening during the generating of the resource map where it appears to be setting the file Name of the resource as well as its line number (code link). To help deep dive into the issue a bit more I modified the code locally which is described by the following code block.

note: forgive my naming of things in advance. I am somewhat confused still about the difference between states and configs.

			if configured {
				var fname string
				ind := fmt.Sprintf("%s.%s", re.ResourceType, re.Name)

				log.Printf("Resource State Type: %s",rs.Type)

				if rs.Type == ResourceTypeData {
					
					log.Println("Resource State Type is Data")
					
					ind = fmt.Sprintf("data.%s", ind)
					fname = filepath.Base(configs[parentConfig].Module.DataResources[ind].Pos.Filename)
					re.Line = &configs[parentConfig].Module.DataResources[ind].Pos.Line
				} else if rs.Type == ResourceTypeResource {
					
					log.Printf("Resource Name: %s",re.Name)
					log.Printf("Resource Type: %s",re.ResourceType)
					for key := range configs[parentConfig].Module.ManagedResources {
						log.Printf("Parent Module Configs Managed Resource Key: %s",key)
					}

					fname = filepath.Base(configs[parentConfig].Module.ManagedResources[ind].Pos.Filename)
					re.Line = &configs[parentConfig].Module.ManagedResources[ind].Pos.Line
				}

				r.AddFileIfNotExists(parent, parentModule, fname)

				parent.Children[fname].Children[id] = re

			}

The following output here shows a bit more whats going on.

2022/01/31 09:48:12 Parent Module State Child Resource: module.atlantis_v2.module.ecs_service.module.alb_routing.data.aws_route53_zone.private
2022/01/31 09:48:12 Resource State Type: data
2022/01/31 09:48:12 Resource State Type is Data
2022/01/31 09:48:12 Parent Module State Child Resource: module.atlantis_v2.module.ecs_service.module.alb_routing.data.aws_route53_zone.public
2022/01/31 09:48:12 Resource State Type: resource
2022/01/31 09:48:12 Resource Name: public
2022/01/31 09:48:12 Resource Type: aws_route53_zone
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_lb_listener_rule.this
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_route53_record.public_record_prevent_destroy
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_route53_record.public_record
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_route53_record.private_record_prevent_destroy
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_route53_record.private_record
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_lb_target_group.this_prevent_destroy
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_lb_target_group.this
2022/01/31 09:48:12 Parent Module Configs Managed Resource Key: aws_lb_listener_rule.this_prevent_destroy

First thing to note here that is worth calling out, there is a 3 depth module call in play. I don't think that matters but maybe it does. It can be seen here at the resource address module.atlantis_v2.module.ecs_service.module.alb_routing.data.aws_route53_zone.public is being labeled as a Resource as opposed to Data. The log further down confirms that the Parent module does not contain a resource for hosted zone. The odd thing here is that the resource prior is very similar, but is treated correctly as a Data type. It has an almost identical path to the impacted resource address.

At the moment I cannot immediately tell how this type is being set because I assume that is where the problem is located or will at least tell us more information about it. FWIW this is a runtime error that is happening when allowing rover to execute the Terraform Plan as well.

I will keep digging into this but if others see anything similar or have some tips help is appreciated.

Thank you for this great tool! It is really useful for a 4-eye plan checking. But it will be really cool to be able to use rover directly with an already existing plan file (json or not) since some organization (as mine) are using some automation and abstraction stuff to generate plans and then apply them. So as a for eye checking it will be very interesting.

Cheers :)

Hi. Thanks for developing this tool. I am not able to run the tool as it tries to generate the TF plan but my configuration uses symlinked files and environment variables to generate the plan. Could you support loading the details directly from a specified plan file instead of generating it on the fly?

As was being discussed in #90, according to @JackFlukinger, there's an issue in the visualization of my AWS-based TF state.

SVG generated by Rover is here: https://file.io/YJrw7hJXmSEp

Let's discuss and fix this issue here instead of in the original bug meant for the panic that got fixed.

Okay @nonbeing , definitely not intended. Where is the data.aws_iam_policy_document.webhook_sns_topic_policy supposed to be housed? Should be a simple fix.

I'm not sure I understand, @JackFlukinger ... Could you please explain what the issue is?

Hi,

This could be by design, but when I go to the Rover - Terraform Visualizer, all the resources are on top of each other in the centre of the graph.

Do I need to move these elements around manually or should it be done by Rover?

Thank you! Dave

Hi, we do have the terraform modules relative to the environment related files. Is it possible to use rover with this setup? /env/env_name/main.tf /env/env_name/state.conf ... configuration.auto.tfvars.... etc /terraform_modules/mod1 /terraform_modules/mod2

Where e.g. main.tf points to module ../../terraform_modules/module1

Thanks

Hi, thanks for that amazing tool.

I found that version v0.3.0 doesn't work while v0.2.2 does work:

docker run --rm -it -p 9000:9000 -v $(pwd):/src --env=AWS_ACCESS_KEY_ID="nanana" --env=AWS_SECRET_ACCESS_KEY="nununu" --env=AWS_DEFAULT_REGION="eu-west-1" --platform linux/amd64 im2nguyen/rover -tfBackendConfig lalala.s3.tfbackend -tfVarsFile lalala.tfvars -workspaceName=lalala
2022/01/14 11:59:40 Starting Rover...
2022/01/14 11:59:40 Initializing Terraform...
2022/01/14 11:59:58 Running in mkyc workspace...
2022/01/14 12:00:03 Generating plan...
2022/01/14 12:00:34 Parsing configuration...
2022/01/14 12:00:34 Generating resource overview...
2022/01/14 12:00:35 Generating resource map...
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x48 pc=0xa5b0cf]

goroutine 1 [running]:
main.(*rover).GenerateModuleMap(0xc000159320, 0xc0003c87e0, {0xc00003d8d8, 0x12})
        /src/map.go:212 +0x1acf
main.(*rover).GenerateModuleMap(0xc000159320, 0xc0006ddb88, {0x0, 0x0})
        /src/map.go:240 +0xefc
main.(*rover).GenerateMap(0xc000159320)
        /src/map.go:319 +0x1bc
main.(*rover).generateAssets(0xc000159320)
        /src/main.go:207 +0x176
main.main()
        /src/main.go:142 +0xd19

when running the same with v0.2.2 it works.

This is a big one.

Adds:

• Full nested module support
• RSO rewritten to support nested modules and spotty configuration
• Graph node and edge refactor to generate from resource map instead of going back through plans, states, configs again
  • Can make changes to graph output by modifying map
• SVG instead of PNG output -- browsers only allow downloaded images up to some unspecified size. I had some trouble with large graphs not exporting due to this. SVG format makes the images much smaller, and infinitely scalable
• Some map updates, always shows line numbers (even in nested modules) if available
  • Nested resources and modules are indented for general readability
• Terraform 1.1.2 support
• SVG export support in CLI with -genImage true. Probably does not work with docker, though.

I'm happy to go into a ton more detail, but it all works. If you don't want to PR because this is so ginormous I get it.

Resolves #60 Resolves #57 Resolves #48 Resolves #46 Resolves #26

Hi,

First of all Rover seems great!

But, I'm trying to use Rover on a CDK for Terraform -project (CDKTF). It is understandable, that Rover does not work in the basedir of the CDK project. So, I cd to the dir where CDKTF creates the Terraform JSON. In that dir what happens is:

% ls
cdk.tf.json     plan
% docker run --rm -it -p 9000:9000 -v $(pwd):/src im2nguyen/rover
2021/09/24 07:50:10 Starting Rover...
2021/09/24 07:50:10 Initializing Terraform...
2021/09/24 07:50:10 Unable to parse Plan: exit status 1

Error: Error building ARM Config: Please ensure you have installed Azure CLI version 2.0.79 or newer. Error parsing json result from the Azure CLI: Error launching Azure CLI: exec: "az": executable file not found in $PATH.


% az --version                                                   
azure-cli                         2.28.0

core                              2.28.0
telemetry                          1.0.6

Python location '/usr/local/Cellar/azure-cli/2.28.0/libexec/bin/python'
Extensions directory '/Users/andreas.asuja/.azure/cliextensions'

Python (Darwin) 3.9.7 (default, Sep  3 2021, 12:37:55) 
[Clang 12.0.5 (clang-1205.0.22.9)]

Legal docs and information: aka.ms/AzureCliLegal


Your CLI is up-to-date.

I'm sorry I don't have time to dig more into this, but just to inform you.

First this tool is awesome!! we use terraform workspaces heavily and would love to have a way to pass the terraform workspace select command some how. Let me know if there is a way!

thanks!

Still have some issue with running rover in my local

How to reproduce

• grab the artifact from releases page
• create main.tf
• run rover_0.3.0_darwin_arm64/rover_v0.3.0 -standalone -tfPath <path-to-terraform-installation>

main.tf ref

output "hello_world" {
  value = "Hello, World!"
}

error log

2022/01/24 14:25:15 Starting Rover...
2022/01/24 14:25:15 Initializing Terraform...
2022/01/24 14:25:15 Generating plan...
2022/01/24 14:25:15 Generating resource overview...
open .terraform/modules/modules.json: no such file or directory
2022/01/24 14:25:15 Generating resource map...
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x2 addr=0xb0 pc=0x10302d520]

goroutine 1 [running]:
main.(*rover).GenerateModuleMap(0x1400000b320, 0x140003c9b68, {0x0, 0x0})
	rover/map.go:103 +0x1e0
main.(*rover).GenerateMap(0x1400000b320)
	rover/map.go:322 +0x1c8
main.(*rover).generateAssets(0x1400000b320)
	rover/main.go:193 +0x10c
main.main()
	rover/main.go:138 +0xeb8

Your tool definitely makes the job! How about some improvements? Here are my suggestions:

• ability to toggle (fold/unfold or hide/unhide) the Legend section
• text box that will help to search for an object
• save as PNG button
• button to reset graph section (IMHO must have)

In terraform you can provide a target module to apply using -target option to only execute a particular module. Does rover support a similar option ? similarly how do you pass the additional options to terraform ?

Hi,

Great tool and benefits are immediately visible. I'm currently using it locally, outside of the CI/CD pipeline as helper before I push the code which brings me to the question of possible scenarios where this can be employed as part of the pipeline. Something like report for the PR approver. Are there such scenarios and what could be the best practice for that kind of integration?

I'm trying to use rover but getting issues with missing variables. This is the command I am using:

docker run --rm -it -p 9000:9000 -v $(pwd):/src im2nguyen/rover -standalone true -tfVar clusterName=eu-cluster

then I see this error:

2022/08/26 20:38:29 Starting Rover...
2022/08/26 20:38:29 Initializing Terraform...
2022/08/26 20:38:36 Generating plan...
2022/08/26 20:38:41 Unable to parse Plan: Unable to run Plan: variable "clusterName" was required but not supplied
[[email protected] terraform]$ docker run --rm -it -p 9000:9000 -v $(pwd):/src im2nguyen/rover -standalone

Is there something I am missing here?

Hi,

I'd suggest Terraform is removed as a dependency in Brew since it can conflict with local setup, e.g. when tfenv is used instead managing Terraform installation with Brew.

Cheers, Alen

I'm trying to build a graphic while running rover (from my terraform project directory) through docker with the following command:

docker run --rm -it -p 9000:9000 -v $(pwd):/src im2nguyen/rover:v0.3.3 -planPath terraform.tfstate

The output would be the following:

2022/08/24 13:54:51 Starting Rover...
2022/08/24 13:54:51 Using provided plan...
2022/08/24 13:54:53 Unable to parse Plan: Unable to read Plan (/src/terraform.tfstate): exit status 1

Error: missing or corrupted provider plugins:
  - registry.terraform.io/hashicorp/random: there is no package for registry.terraform.io/hashicorp/random 3.3.2 cached in .terraform/providers
  - registry.terraform.io/hashicorp/time: there is no package for registry.terraform.io/hashicorp/time 0.7.2 cached in .terraform/providers
  - registry.terraform.io/hashicorp/cloudinit: there is no package for registry.terraform.io/hashicorp/cloudinit 2.2.0 cached in .terraform/providers
  - registry.terraform.io/gavinbunney/kubectl: there is no package for registry.terraform.io/gavinbunney/kubectl 1.14.0 cached in .terraform/providers
  - registry.terraform.io/hashicorp/external: there is no package for registry.terraform.io/hashicorp/external 2.2.2 cached in .terraform/providers
  - registry.terraform.io/hashicorp/aws: there is no package for registry.terraform.io/hashicorp/aws 3.75.2 cached in .terraform/providers
  - registry.terraform.io/hashicorp/null: there is no package for registry.terraform.io/hashicorp/null 3.1.1 cached in .terraform/providers
  - registry.terraform.io/hashicorp/tls: there is no package for registry.terraform.io/hashicorp/tls 4.0.1 cached in .terraform/providers
  - registry.terraform.io/hashicorp/helm: there is no package for registry.terraform.io/hashicorp/helm 2.6.0 cached in .terraform/providers
  - registry.terraform.io/hashicorp/kubernetes: there is no package for registry.terraform.io/hashicorp/kubernetes 2.12.1 cached in .terraform/providers
  - registry.terraform.io/hashicorp/http: there is no package for registry.terraform.io/hashicorp/http 3.0.1 cached in .terraform/providers
  - registry.terraform.io/hashicorp/local: there is no package for registry.terraform.io/hashicorp/local 2.2.3 cached in .terraform/providers

I can however start rover through passing the plan as a json:

docker run --rm -it -p 9000:9000 -v $(pwd):/src im2nguyen/rover:v0.3.3 -planJSONPath=plan.json

If I access the container via CLI I can see that the terraform version used is 1.1.2 when the version used for the deployment is 1.1.9. I've tried passing the terraform path through -tfPath hoping it would take the correct one used for my deployment but to no help.

How can I fix the versioning issue so that terraform init doesn't fail? (I'm also parsing it as a json to work because if I don't parse any plan rover will freeze during terraform init, most likely due to this issue)

Also, is there a way to generate a graphic (since I want to include it in a pipeline) with only specific resources shown, like for example only created and updated resources without anything else (especially variables) ?

Thanks in advance!