Coder.com•13mo ago

How does the coder_script terraform resource work?

I have a script that I need to run when the workspace has stopped/deleted to delete some resources from another k8s cluster. I add it as a resource to the template and set “run_on_stop” to true and set the agent id to the one of the workspace but when I stop/delete a workspace it doesn’t run. Am I missing something in the setup?

30 Replies

Codercord•13mo ago

<#1210530942246985740>

Category

Help needed

Product

Coder OSS (v2)

Platform

Linux

Logs

Please post any relevant logs/error messages.

Atif•13mo ago

This only runs on stop transition. i.e. when you press the stop button and not on deleting it. Each start/stop/update/create are actually terraform apply with the difference of data.coder_workspace.me.start_count being 0 for the stop case. And the delete operation is terraform destroy So it will just delete all resources. You should be able to run a script at stop transition when you use run_at_stop = true

ConorOP•13mo ago

So I have the script with run_on_stop set to true and when I press stop on the workspace it doesn’t run. Also would there be a way to run a script before the deletion of a workspace?

kyle•13mo ago

For that I'd recommend using the local-exec resource instead.

Codercord•13mo ago

@Phorcys closed the thread.

mingfang•2mo ago

I have the same use case; I need to manually stop some external processes on shutdown. How can local-exec help in this case?

Phorcys•2mo ago

hey @mingfang, local-exec runs a command on the provisioner (not within the workspace), you can use it to run a script on workspace deletion you can do something like this

resource "coder_agent" "dev" {
  arch = "amd64"
  os   = "linux"
  dir  = "/workspace"

  provisioner "local-exec" {
     when    = "destroy"
     command = "<do something>"
  }
}

resource "coder_agent" "dev" {
  arch = "amd64"
  os   = "linux"
  dir  = "/workspace"

  provisioner "local-exec" {
     when    = "destroy"
     command = "<do something>"
  }
}

see https://developer.hashicorp.com/terraform/language/resources/provisioners/local-exec#example-usage for more info let me know if you need anything else

mingfang•2mo ago

@Phorcys thanks for the explanation. for my use case I need the script to run on the workspace. i was hoping that I can use coder_script https://registry.terraform.io/providers/coder/coder/latest/docs/resources/script#run_on_stop-1 but i have not been able to get a simple echo hello to work

Phorcys•2mo ago

sadly you can't have the script run on deletion yoh can have it run on start or stop only if a workspace is not started up and gets deleted, it only gets deleted and not started up, so your scripts can't run inside of it may I as what the use-case is?

Atif•2mo ago

This is a good feature request, we can invetsigate to add to the provider. Coul you open a feature request issue on GitHub?

Phorcys•2mo ago

is it technically feasible though? a delete script won't be able to run inside the workspace because the workspace would have to be running for it to work

Atif•2mo ago

We can run it just before deleting. Similar to how a script can run on stop. Take for example the case of docker template where the compute(container) gets deleted on stop but can still run the script before getting destroyed. Makes sense?

Phorcys•2mo ago

i know but what if the workspace is deleted when stopped

mingfang•2mo ago

Sorry I was not being clear. I'm trying to run a script on stop and not on delete. I've tried these and they are not working.

resource "coder_agent" "pod" {
  shutdown_script = <<-EOT
    echo "this does not get run on stop"
    EOT
}

resource "coder_agent" "pod" {
  shutdown_script = <<-EOT
    echo "this does not get run on stop"
    EOT
}

and

resource "coder_script" "shutdown" {
  agent_id     = coder_agent.pod.id
  run_on_stop  = true
  script       = <<-EOF
    echo "this does not get run on stop"
    EOF
}

resource "coder_script" "shutdown" {
  agent_id     = coder_agent.pod.id
  run_on_stop  = true
  script       = <<-EOF
    echo "this does not get run on stop"
    EOF
}

Question is...why is this not working?

Phorcys•2mo ago

good question, could you send your full template over please?

mingfang•2mo ago

My template is made up of several Terraform files. https://github.com/mingfang/coder-templates/tree/master/catalog The shutdow script is here https://github.com/mingfang/coder-templates/blob/master/catalog/main.tf#L108 The purpose of this template is to provide a catalog of services to try out. The service should start and shutdown along of the workspace. You can click on the link to the service(e.g. Kestra) from github to try it yourself.

GitHub

coder-templates/catalog at master · mingfang/coder-templates

Contribute to mingfang/coder-templates development by creating an account on GitHub.

GitHub

coder-templates/catalog/main.tf at master · mingfang/coder-templates

Contribute to mingfang/coder-templates development by creating an account on GitHub.

Phorcys•2mo ago

can you check the version of the coder/coder Terraform provider? it should be in .terraform.lock.hcl or maybe you can find it via the TF CLI

mingfang•2mo ago

https://github.com/mingfang/coder-templates/blob/master/catalog/providers.tf I left it blank so it should be the "latest", correct?

GitHub

coder-templates/catalog/providers.tf at master · mingfang/coder-tem...

Contribute to mingfang/coder-templates development by creating an account on GitHub.

Phorcys•2mo ago

I saw this, if you do not have a .terraform.lock.hcl lockfile then yes, it'll be latest

mingfang•2mo ago

I don't have a lock file. Here is the build log

Initializing provider plugins...
- Finding latest version of hashicorp/kubernetes...
- Finding coder/coder versions matching ">= 0.17.0"...
- Installing hashicorp/kubernetes v2.35.1...
- Installed hashicorp/kubernetes v2.35.1 (signed by HashiCorp)
- Installing coder/coder v2.1.2...
- Installed coder/coder v2.1.2 (signed by a HashiCorp partner, key ID 93C75807601AA0EC)
Partner and community providers are signed by their developers.
If you'd like to know more about provider signing, you can read about it here:
https://www.terraform.io/docs/cli/plugins/signing.html
Terraform has created a lock file .terraform.lock.hcl to record the provider
selections it made above. Include this file in your version control repository
so that Terraform can guarantee to make the same selections by default when
you run "terraform init" in the future.
Terraform has been successfully initialized!

Initializing provider plugins...
- Finding latest version of hashicorp/kubernetes...
- Finding coder/coder versions matching ">= 0.17.0"...
- Installing hashicorp/kubernetes v2.35.1...
- Installed hashicorp/kubernetes v2.35.1 (signed by HashiCorp)
- Installing coder/coder v2.1.2...
- Installed coder/coder v2.1.2 (signed by a HashiCorp partner, key ID 93C75807601AA0EC)
Partner and community providers are signed by their developers.
If you'd like to know more about provider signing, you can read about it here:
https://www.terraform.io/docs/cli/plugins/signing.html
Terraform has created a lock file .terraform.lock.hcl to record the provider
selections it made above. Include this file in your version control repository
so that Terraform can guarantee to make the same selections by default when
you run "terraform init" in the future.
Terraform has been successfully initialized!

Phorcys•2mo ago

thanks this is indeed the latest i will take a deeper look in a moment, this behavior from the provider is pretty weird

mingfang•2mo ago

It turns out the shutdown script is running but it didn't get a chance to finish. I'm guessing it is because the ssh server got killed before the script gets a chance to finish. Here is the part of the agent log that shows the shutdown sequence.

2025-01-31 09:21:29.701 [info]  shutting down agent
2025-01-31 09:21:29.701 [debu]  set lifecycle state  current={"state":"shutting_down","changed_at":"2025-01-31T09:21:29.701491Z"}  last={"state":"ready","changed_at":"2025-01-31T09:18:49.701169Z"}
2025-01-31 09:21:29.701 [info]  ssh-server: stopped serving listener  listen_addr={} ...
2025-01-31 09:21:29.701 [info]  running agent script  log_source_id=ac47a7e3-34a9-44e6-8fdd-b94a9ec8cc5d  log_path=/home/coder/shutdow.log  script_data_dir=/tmp/coder-script-data/ac47a7e3-34a9-44e6-8fdd-b94a9ec8cc5d ...
    script= #!/bin/sh
            cd "/home/coder/repo/examples/postgrest"
            terraform destroy -auto-approve
            echo "bye bye bye"
2025-01-31 09:21:30.196 [info]  connecting to coderd
2025-01-31 09:21:30.200 [warn]  run exited with error ...

2025-01-31 09:21:29.701 [info]  shutting down agent
2025-01-31 09:21:29.701 [debu]  set lifecycle state  current={"state":"shutting_down","changed_at":"2025-01-31T09:21:29.701491Z"}  last={"state":"ready","changed_at":"2025-01-31T09:18:49.701169Z"}
2025-01-31 09:21:29.701 [info]  ssh-server: stopped serving listener  listen_addr={} ...
2025-01-31 09:21:29.701 [info]  running agent script  log_source_id=ac47a7e3-34a9-44e6-8fdd-b94a9ec8cc5d  log_path=/home/coder/shutdow.log  script_data_dir=/tmp/coder-script-data/ac47a7e3-34a9-44e6-8fdd-b94a9ec8cc5d ...
    script= #!/bin/sh
            cd "/home/coder/repo/examples/postgrest"
            terraform destroy -auto-approve
            echo "bye bye bye"
2025-01-31 09:21:30.196 [info]  connecting to coderd
2025-01-31 09:21:30.200 [warn]  run exited with error ...

Phorcys•2mo ago

oh yeah that makes sense what if you add start_blocks_login = true to the coder_script?

mingfang•2mo ago

that caused a validation error

Error: "start_blocks_login" can only be set if "run_on_start" is "true"

Error: "start_blocks_login" can only be set if "run_on_start" is "true"

mingfang•2mo ago

should this https://github.com/coder/coder/blame/f6e990ed875f7e7f273b8a6f5c03c3475a7bb026/agent/agent.go#L1552 run before the ssh server shutdown here https://github.com/coder/coder/blame/f6e990ed875f7e7f273b8a6f5c03c3475a7bb026/agent/agent.go#L1537 ?

GitHub

Blaming coder/agent/agent.go at f6e990ed875f7e7f273b8a6f5c03c3475a7...

Provision remote development environments via Terraform - Blaming coder/agent/agent.go at f6e990ed875f7e7f273b8a6f5c03c3475a7bb026 · coder/coder

Phorcys•2mo ago

hey @mingfang, have you figured this out, or is this issue still happening?

mingfang•2mo ago

it's still a problem. I think the coder above may be the problem; i think the stop scripts should run before the ssh server is shutdown

Phorcys•7d ago

@mingfang apologies for not getting back to you sooner, would you be able to open up an issue on our GitHub?

Phorcys•7d ago

https://github.com/coder/coder/issues

GitHub

Issues · coder/coder

Provision remote development environments via Terraform - Issues · coder/coder

Phorcys•7d ago

make sure to mention the Coder version and your terraform provider version as well as the analysis you've provided

Gaming

Programming

How does the coder_script terraform resource work?

Did you find this page helpful?