TL;DR

This project makes use of the null_data_source datas, which is deprecated in favor of locals (docs). This trigger warnings

Expected behavior

No warnings are triggered when using this module

Observed behavior

Warnings are triggered when using this module

Terraform Configuration

Any configuration should trigger the warnings

Terraform Version

1.5.4

Additional information

I have stumbled upon this while using https://github.com/terraform-google-modules/terraform-google-scheduled-function

This is not a bug per se, but I think it's the most appropriate template

Here are the 2 deprecated null_data_source:

• terraform-google-event-function/examples/automatic-labelling-from-repository/main.tf

  Line 42 in 8d977a4

  data "null_data_source" "main" {

• terraform-google-event-function/main.tf

  Line 42 in 8d977a4

  data "null_data_source" "wait_for_files" {

The terraform resource supports a vpc_connector . It would be great to include it in the module as well. We've had to make a custom change to use this in our environment. Would be great to avoid drift from the original module.

TL;DR

whenever I run apply for google_cloudfunctions_function it always re-creates the function. Even if I've made no changes to config.

Expected behavior

The function should remain untouched

Observed behavior

function is destroyed and re-created

  # google_cloudfunctions_function.export-surveys-function is tainted, so must be replaced
-/+ resource "google_cloudfunctions_function" "export-surveys-function" {
      - environment_variables         = {} -> null
      + https_trigger_url             = (known after apply)
      ~ id                            = "projects/project/locations/europe-west1/functions/function-name" -> (known after apply)
      - labels                        = {} -> null
        name                          = "function-name"
      + vpc_connector_egress_settings = (known after apply)
        # (15 unchanged attributes hidden)

        # (5 unchanged blocks hidden)
    }

Terraform Configuration

resource "google_cloudfunctions_function" "export-surveys-function" {
  docker_registry       = "CONTAINER_REGISTRY"
  name                  = local.function_name
  description           = "Function that is triggered by a schedule (via pubsub) to export all survey data from mongo"
  region                = var.region
  entry_point           = "entryPoint"
  source_archive_bucket = google_storage_bucket.source_code_bucket.name
  source_archive_object = "function.zip"
  runtime               = "nodejs16"
  labels                = {}
  project = data.google_project.project.project_id
  service_account_email  = data.google_service_account.run-account.email
  min_instances         = 0
  environment_variables = {}
  https_trigger_security_level  = "SECURE_ALWAYS"
}

Terraform Version

Terraform v1.3.1
on darwin_arm64
+ provider registry.terraform.io/hashicorp/google v4.38.0

Additional information

No response

To make the examples more accessible to users, they should not rely on a hard-coded network name.

As described here.

Hey folks,

It would be good to module support the excludes parameter in arquive_file step.

From:

data "archive_file" "main" {
  type        = "zip"
  output_path = pathexpand("${var.source_directory}.zip")
  source_dir  = data.null_data_source.wait_for_files.outputs["source_dir"]
}

To:

data "archive_file" "main" {
  type        = "zip"
  output_path = pathexpand("${var.source_directory}.zip")
  source_dir  = data.null_data_source.wait_for_files.outputs["source_dir"]
  excludes     = var.files_to_exclude_in_source_dir
}

TL;DR

Currently all functions deployed via TF are deployed using the legacy v1 of the run-time. There is no way to create a gen 2 function

Terraform Resources

google_cloudfunctions_function - https://registry.terraform.io/providers/hashicorp/google/latest/docs/resources/cloudfunctions_function

Detailed design

Information on the differences between v1 and v2 here

Additional information

No response

TL;DR

Given a function that's deployed with a source_archive_bucket when the bucket is updated by terraform it does not re-deploy the function.

Expected behavior

Changing the contents of the source code bucket should re-deploy the function to update the source.

Observed behavior

the source is not updated, The work-around is that you have to do: terraform plan -replace= to force TF to replace the function every time.

Terraform Configuration

resource "google_storage_bucket" "source_code_bucket" {
...
}

resource "google_storage_bucket_object" "function" {
  name   = "function.zip"
  source = "../dist/function.zip"
  bucket = google_storage_bucket.source_code_bucket.name
}

resource "google_cloudfunctions_function" "export-surveys-function" {
...
  source_archive_bucket = google_storage_bucket.source_code_bucket.name
  source_archive_object = "function.zip"
 }

Terraform Version

Terraform v1.2.8
on darwin_arm64
+ provider registry.terraform.io/hashicorp/google v4.38.0

Additional information

No response

When I modify a file in the source_directory, I find that the zip file is not recreated when one of the files in the source directory has changed.

It seems we fixed a part of this when fixing #32 with PR #35 by forcing update on the object in the GCS bucket - but the archive_file (zip) itself is not being updated, i.e in the end not forcing the update of the Cloud Function.

The upstream issue for failing to update the archive_file is here: hashicorp/terraform-provider-archive#39. It seems it's linked to running plan and apply in a different context (think a Cloud Build pipeline composed of two configs that effectively 'forgets' about the .terraform/ folder between plan and apply).

I see 3 ways we can work-around this in this module (while waiting for an upstream fix):

• compute the hash of all files in the source directory and force the recomputing of the archive_file resource based on if any hash in the source_directory has changed; but I'm not certain how to do this for all files in the source_directory (for one file, it would be straightforward - for multiple, we'll need a custom shell script to list out all the files in the directory).

• have a random id appended to the zip filename that would be forced to regenerate when we need it, i.e by passing e.g a force_update variable to the module.

• advertise that plan and apply should be run in the same context, which would make some users struggle.

Any help is appreciated here :)

It would be nice to somehow indicate in which subdirectory the source zip file should be placed.

For example, I have a single bucket for all cloud functions sources, and I want to keep a file layout like this:

my-bucket/
    function1/hash-build.zip
    function2/hash-build.zip

My suggestion is to add an optional variable for object directory path.

I have noticed that using for_each with this module will cause inconsistent results because the source folder is overwritten by the successive executions in the for_each loop. In general, we should avoid creating local directories to build GCF code until Terraform accepts temp directories.

To fix this, imo it is better to not define any local directory and specify the source for files directly since we don't manipulate local files anymore. It will also help to get rid of the null_resource.wait_for_files resource hack.

We should switch from this:

data "archive_file" "main" {
  type        = "zip"
  output_path = pathexpand("${var.source_directory}.zip")
  source_dir  = "${data.null_data_source.wait_for_files.outputs["source_dir"]}"
}

to

resource "random_uuid" "this" {
  keepers = {
    for filename in fileset(local.function_source_directory, "**/*") :
    filename => filemd5("${local.function_source_directory}/${filename}")
  }
}

data "archive_file" "main" {
  type        = "zip"
  output_path = pathexpand("${var.source_directory}.zip")
  dynamic "source" {
    for_each = var.files
    content {
      content = each.value.content
      filename = each.value.filename
    }
  }
}

and switch inputs to

files = [ 
 { content = "test", filename = "test.txt" },
 { content = "test2", filename = "test2.txt" } 
]

See terraform-google-modules/terraform-google-slo#38 (comment)

Relevant issues:
terraform-google-modules/terraform-google-vpc-service-controls#11
#27

We need to have an event function for automatic project labeling in a folder.

The terraform resource supports a vpc_connector and vpc_connector_egress_settings. It would be great to include it in the module as well. We've had to make a custom change to use this in our environment. Would be great to avoid drift from the original module.

TL;DR

We are seeing some flakes in creating Cloud Functions with a generic "Failed to retrieve function source code" error message. Can we improve the error message to indicate what is wrong?

Expected behavior

Succeed or output a message that is useful to troubleshoot.

Observed behavior

The message is too generic. The same deployment has passed before, and we don't know what is causing this flake.

Terraform Configuration

module "batch_cf" {
  source                = "terraform-google-modules/event-function/google//modules/repository-function"
  version               = "2.1.0"
  name                  = "batch-function"
  project_id            = module.project.project_id
  description           = "Cloud Function to initiate batch pipeline"
  timeout_s             = 540
  entry_point           = "launch_batch_pipeline"
  service_account_email = "${google_service_account.function_runner.account_id}@${module.project.project_id}.iam.gserviceaccount.com"

  source_repository_url = "https://source.developers.google.com/projects/<myproject>/repos/<myrepo>/fixed-aliases/v1.0.0"
  runtime               = "python37"

  region = "us-central1"

  event_trigger = {
    event_type                         = "google.storage.object.finalize"
    resource                           = "<mybucketname>"
    event_trigger_failure_policy_retry = true
  }

  environment_variables = {
    LOCATION                   = "us-central1"
    TEMPLATE_FILE_GCS_LOCATION = "gs://<path to file>.json"
  }
}

Terraform Version

0.14.8

Additional information

No response

Terraform is giving off a warning which fails some CI checks when running terraform plan via GitHub actions. We no longer need to specify ${}for expressions in >=12 .

I'll create a PR to address this shortly but wanted an issue first.

TL;DR

We're checking for null, so setting it to false will result in an error.

Expected behavior

If trigger_http = false and event_trigger is defined, the module should work.

Observed behavior

Error message:

One of event_trigger or trigger_http is required: You must specify a trigger when deploying a new function.

Terraform Configuration

module "cloud-function" {
  source              = "terraform-google-modules/event-function/google"
  version             = "2.5.0"
  name                = "name"
  project_id          = var.project_id
  region              = var.region
  runtime             = "nodejs16"
  timeout_s           = 540  # 9min, maximum for gen1 cloud functions (gen2 will have 3600)
  available_memory_mb = 512
  bucket_name         = "${var.project_id}-name"
  source_directory    = "source"
  entry_point         = "handleMessage"
  trigger_http        = false
  event_trigger = {
    event_type = "google.pubsub.topic.publish"
    resource   = "projects/project/topics/topic"
  }
}

Terraform Version

v1.3.4

Additional information

No response

Relevant issue: terraform-google-modules/terraform-google-vpc-service-controls#11

If I am using the local_directory event-function I'd expect modifying the files in that directory (and running terraform apply) to redeploy the function.

module "localhost_function" {
  source  = "terraform-google-modules/event-function/google"
  version = "1.1.0"

  description = "Adds projects to VPC service permiterer."
  entry_point = "handler"

  environment_variables = {
    FOLDER_ID = var.folder_id
  }

  event_trigger    = module.event_folder_log_entry.function_event_trigger
  name             = local.watcher_name
  project_id       = var.project_id
  region           = var.region
  source_directory = abspath(path.module)
  runtime          = "python37"
  //  runtime               = "go111"
  available_memory_mb   = 2048
  timeout_s             = 540
  service_account_email = google_service_account.watcher.email
}

Currently it doesn't update the function if I do this. I have to explicitly recreate the function.

We should update this so changes to the source directory force a function reupload.

TL;DR

https://www.terraform.io/language/resources/syntax#operation-timeouts

Add timeout block to customize how long certain operations are allowed to take before being considered to have failed.
https://registry.terraform.io/providers/hashicorp/google/latest/docs/resources/cloudfunctions_function#timeouts

Terraform Resources

No response

Detailed design

No response

Additional information

No response

TL;DR

The google_storage_bucket_object.main must be replaced if we apply first on Linux then on Mac (or conversly). This comes from the md5 of the archive_file data being different between the two operating systems because the generated zip file does not have the same system permission, 0664 on linux and 0644 on macOS.

Expected behavior

The resource google_storage_bucket_object.main should not diverge between two operating systems. The output_file_mode field on the data archive_file should be a defaulted variable to avoid this issue.

Observed behavior

No response

Terraform Configuration

module "function" {
  source  = "terraform-google-modules/event-function/google"
  version = "~> 2.3.0"

  entry_point = "ProcessPubSub"
  event_trigger = {
    event_type = "google.pubsub.topic.publish"
    resource   = module.pubsub_topic.id
  }
  name             = "${local.name}-exporter"
  project_id       = var.project
  region           = var.region
  runtime          = "go116"
  source_directory = "${path.module}/function"

  source_dependent_files = []

  available_memory_mb                = 256
  bucket_force_destroy               = true
  bucket_name                        = "${local.name}-scheduled-exporter-function-${random_id.suffix.hex}"
  description                        = "Function to backup SQL instance."
  event_trigger_failure_policy_retry = false
  service_account_email              = google_service_account.this.email
  timeout_s                          = 540
}

### Terraform Version

```sh
1.3.0

Additional information

No response

TL;DR

Pushing the source code to storage using ADO pipeline, Referring the object generation in the cloud function.

Expected behavior

Re-deploying the cloud function should update the source code.

Observed behavior

Re-deploying the cloud function can deploy the latest version, but source code is not updated.

Terraform Configuration

data "google_storage_bucket_object" "abc" {
  name   = var.abc
  bucket = data.google_storage_bucket.bucket.name
}

resource "google_cloudfunctions_function" "abc" {
  project               = var.project
  name                  = var.abc
  region                = var.region
  description           = var.description
  runtime               = var.runtime
  service_account_email = module.service_accounts.email

  available_memory_mb   = 128
  source_archive_bucket = data.google_storage_bucket_object.abc.bucket
  source_archive_object = data.google_storage_bucket_object.abc.name
  entry_point           = var.entrypoint

  event_trigger {
    event_type = var.triggertype
    resource   = var.resource
  }

  environment_variables = {
    object_url = data.google_storage_bucket_object.abc.media_link
  }
}

Terraform Version

1.4.2

Additional information

No response

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Pending Status Checks

These updates await pending status checks. To force their creation now, click the checkbox below.

• chore(deps): Update go modules and/or dev-tools (cloud.google.com/go, google.golang.org/api)
• chore(deps): Update Terraform terraform-google-modules/event-function/google to v4
• chore(deps): Update Terraform terraform-google-modules/project-factory/google to v16
• chore(deps): Update dependency googleapis to v142

Detected dependencies

• Check this box to trigger a request for Renovate to run again on this repository

I am trying to create a Cloud Function using the terraform module. My configuration is as follows:

module "test_function" {
  source      = "terraform-google-modules/event-function/google//modules/repository-function"
  version     = "1.4.0"
  description = "Cloud Function for Testing"
  entry_point = "Process"
  event_trigger = {
    event_type = "google.storage.object.finalize"
    resource   = "projects/{{.my_test_project}}/buckets/test-bucket"
  }
  name                  = "test-function"
  project_id            = "{{.my_test_project}}"
  region                = "us-central1"
  source_repository_url = "https://source.developers.google.com/projects/{{.my_test_project}}/repos/{{.my-repo}}/main"
  environment_variables = {
    GOOGLE_CLOUD_PROJECT       = "{{.my_test_project}}"
    REGION                     = "us-central1"
  }
  timeout_s = 60
  runtime   = "go111"
}

Running terraform init and terraform plan completes successfully, but terraform apply returns

Error: googleapi: Error 400: The request has errors, badRequest

on .terraform/modules/test_function/modules/repository-function/main.tf line 17, in resource "google_cloudfunctions_function" "main":
17: resource "google_cloudfunctions_function" "main" {

I can't tell from the error message what is wrong and the configuration seems fine. A quick search showed other people running into the same problem but their solutions have not worked for me. Are there suggestions for what I'm missing or how to find a more detailed error message?

Thanks!

This PR:
#23

References @aaron-lane TF 0.12 branch, it should be updated after new version is released.

Both make check and make docker_verify fail

TL;DR

Use of this module generates a warning about the use of the deprecated null_data_source.

Expected behavior

No warnings by Terraform.

Observed behavior

The following warning is shown when using the module:

Warning: Deprecated Resource
 with module.scheduled-function.module.main.data.null_data_source.wait_for_files,
 on .terraform/modules/scheduled-function.main/main.tf line 32, in data "null_data_source" "wait_for_files":
 32: data "null_data_source" "wait_for_files" {
The null_data_source was historically used to construct intermediate values
to re-use elsewhere in configuration, the same can now be achieved using
locals

Terraform Configuration

n/a

Terraform Version

Terraform v1.2.7
on darwin_arm64

Additional information

No response

Support the argument source_archive_bucket to enable passing a zip in a bucket already created by previous TF code.
This will unlock 0.13 upgrade as well, since currently looping through this module will result in GCF source code (shared folder) being overwritten by each for_each element.

We are currently only supporting project-level log exports. Log exports can also be aggregated at the folder or organization level: it would be nice to support them.

Suggest using the terraform-google-log-export module that already supports all 3 log exports, and soon will support billing-level log exports as well.

Additionally, sometimes users want to export logs from the GSuite Admin APIs: we recommend using the terraform-google-gsuite-export for this use case.