Hi there!
Since few days I had this error message when running terraform plan:
[...]
element "XXX": attribute "XXX": element "XXX": object required.
I dug around and fell on this thread: The given value is not suitable for module.
It took me a moment to understand great explanations of master @apparentlymart , but at the end, I make it!
For those who wants an illustration to understand and fix this object required issue:
Here is a working code:
virtual_networks = {
test = {
division = "test",
prefix = ["10.2.71.0/24"],
int_id = "c2",
subnets = {
test = {
prefix = ["10.2.71.0/25"],
env = "test",
application_name = "xxx",
application_layout = "development",
route_table = "standard",
int_id = "c2"
},
test_pe = {
prefix = ["10.2.71.128/25"],
env = "test",
application_name = "xxx",
application_layout = "developmentpe",
route_table = "standard",
int_id = "c2"
}
}
}
}
Here is the corresponding tfvars:
variable "virtual_networks" {
description = "Map of virtual networks to create with their subnets"
type = map(object({
division = string,
prefix = list(string),
int_id = optional(string),
subnets = map(object({
env = string,
application_name = string,
application_layout = string,
int_id = optional(string),
prefix = list(string)
route_table = string
}))
}
)
)
}
What is important here is to notice **that** 
:
[...]
subnets = {
**test = {**
prefix = ["10.2.71.0/25"]
[...]
},
**test_pe = {**
prefix = ["10.2.71.128/25"]
[...]
}
}
Subnet is effectively referencing 2 maps: test and test_pe.
Now, here is a NOT working code:
test_compute_clusters = {
mlcc_test = {
index = "1",
division = "devpe",
identity_type = "SystemAssigned",
priority = "LowPriority",
node_public_ip_enabled = false,
virtual_machine_size = "STANDARD_DS4_V2",
test_workspace_key = "test",
subnet_key = "snet-001",
scale_settings = {
max_node_count = 1,
min_node_count = 0,
scale_down_nodes_after_idle_duration = "PT3M"
}
}
With this tfvars:
variable "test_compute_clusters " {
description = "Machine learning compute cluster definition"
type = map(object({
index = optional(string),
context = string,
identity_type = string,
priority = string,
node_public_ip_enabled = optional(bool, false),
virtual_machine_size = string,
test_workspace_key = string,
subnet_key = string,
scale_settings = map(object({
max_node_count = number,
min_node_count = number,
scale_down_nodes_after_idle_duration = string
}))
}))
}
What is important here is to notice ** that ** :
subnet_key = "snet-001",
scale_settings = {
**max_node_count = 1,**
**min_node_count = 0,**
**scale_down_nodes_after_idle_duration = "PT3M"**
}))
scale_settings is NOT REFERENCING a map! but objects only!!!
If I really want scale_settings being a map and avoid the object required message, I should had written this:
subnet_key = "snet-001",
scale_settings = {
settings_1 = { #<============ HERE IS MY MISTAKE
max_node_count = 1,
min_node_count = 0,
scale_down_nodes_after_idle_duration = "PT3M"
}}))
and not this
subnet_key = "snet-001",
scale_settings = {
max_node_count = 1, #<===== SEE HERE? NO MAP BEFORE `max_node_count`!
min_node_count = 0,
scale_down_nodes_after_idle_duration = "PT3M"
}}))
So, Terraform was correct, what I needed here was an object and not a map of object!
I hope this post will be useful to anybody & Thanks to @apparentlymart !
Have a nice day!