database creation stuck in configuration phase

[hFti]

Affected modules

terraform-ibm-modules/icd-postgresql/ibm
version 3.17.9

Terraform CLI and Terraform provider versions

• Terraform version: 1.5.7
• Provider version: 1.68.1

Terraform output

Expected behavior

databse created with conf in place

Actual behavior

configuration is stuck , terraform timed out and the database in console is always at 60%

Steps to reproduce (including links and screen captures)

use module with configuration variable set to the map below
`module "postgresql_db" {
source = "terraform-ibm-modules/icd-postgresql/ibm"
version = "3.17.9"
resource_group_id = ibm_resource_group.resource_group.id
name = "${var.env}-database-postgres"
region = var.region
pg_version = var.pg_version
admin_pass = var.admin_pass
users = var.users
kms_encryption_enabled = true
kms_key_crn = module.key_protect_all_inclusive.keys["icd-key"].crn
existing_kms_instance_guid = module.key_protect_all_inclusive.kms_guid
resource_tags = var.resource_tags
service_credential_names = var.service_credential_names
access_tags = var.access_tags
member_host_flavor = "b3c.4x16.encrypted"
auto_scaling = var.auto_scaling

configuration = {
shared_buffers = 32000
max_connections = 250
max_locks_per_transaction = 64
max_prepared_transactions = 0
synchronous_commit = "local"
effective_io_concurrency = 12
deadlock_timeout = 10000
log_connections = "off"
log_disconnections = "off"
log_min_duration_statement = 100
tcp_keepalives_idle = 200
tcp_keepalives_interval = 50
tcp_keepalives_count = 6
archive_timeout = 1000
wal_level = "hot_standby"
max_replication_slots = 10
max_wal_senders = 20
}
cbr_rules = [
{
description = "${var.env}-postgres access only from vpc"
enforcement_mode = "enabled"
account_id = data.ibm_iam_account_settings.iam_account_settings.account_id
rule_contexts = [{
attributes = [
{
"name" : "endpointType",
"value" : "private"
},
{
name = "networkZoneId"
value = module.cbr_zone.zone_id
}]
}]
}
]
}`
run terraform apply

Anything else

[hFti]

@lionelmace

[ocofaigh]

@hFti I suggest you create a support case with IBM Cloud for this. This has nothing to do with our module. It also looks like you might have escaped the apply early (maybe you hit ctrl + c) ? This is never a good idea in terraform. You should always allow terraform to timeout itself or it can corrupt your state file.

[lionelmace]

I tried and the terraform timed out after 1h10mins

module.postgresql_db.ibm_database.postgresql_db: Still creating... [1h11m20s elapsed]
module.postgresql_db.ibm_database.postgresql_db: Still creating... [1h11m30s elapsed]
╷
│ Error: [ERROR] Error waiting for database (crn:v1:bluemix:public:databases-for-postgresql:eu-de:a%2Fb41a28b85f604bf6b86986adb5117b13:fa9e04bb-f9be-4262-8182-d977d87aa22e::) configuration update task to complete: [Error] Time out waiting for database task to complete
│
│   with module.postgresql_db.ibm_database.postgresql_db,
│   on .terraform/modules/postgresql_db/main.tf line 55, in resource "ibm_database" "postgresql_db":
│   55: resource "ibm_database" "postgresql_db" {

Applying new configuration also hangs at 60%.

Here are the terraform scripts I used to test: https://github.com/lionelmace/ibmcloud-utils/tree/master/terraform/icd-postgres-modules

@hFti I agree with @ocofaigh . Please create a support case.

[lionelmace]

@ocofaigh The example in the module is incorrect. It should be fixed to reflect the following changes

It's failing as logical replication is not enabled . "wal_level" must be defined properly .

Make sure that your database is configured such that logical replication can be enabled. This means the wal_level needs to be set to logical. Also, max_replication_slots and max_wal_senders must be greater than 20.

You can refer below document too.
 https://registry.terraform.io/providers/IBM-Cloud/ibm/latest/docs/resources/database

The following example does work
https://github.com/lionelmace/ibmcloud-utils/blob/master/terraform/icd-postgres-modules/database-postgres.tf

[ocofaigh]

@lionelmace yea there was an ICD backend schema bug. The old schema allowed hot_standby as a value and didn't allow replica. However the new schema now allows replica (which is the default) and doesn't allow hot_standby. We will update the example now

[ocofaigh]

@lionelmace Actually it was fixed already in #514