upcloud_managed_database_postgresql complains about properties

[jylitalo]

provider: upcloudltd/upcloud v5.1.1

terraform apply/refresh/plan produces:

│ Error: Invalid address to set: []string{"properties", "0", "pg_read_replica"}
│
│   with upcloud_managed_database_postgresql.db,
│   on main.tf line 30, in resource "upcloud_managed_database_postgresql" "db":
│   30: resource "upcloud_managed_database_postgresql" "db" {

terraform in question:

resource "upcloud_managed_database_postgresql" "db" {
  name  = "gateway-db-${var.env}"
  plan  = var.db_plan
  title = "gateway-db-${var.env}"
  zone  = data.terraform_remote_state.network.outputs.zone
  properties {
    admin_username = var.worker_db_username
    admin_password = random_password.worker_db_password.result
    ip_filter      = data.terraform_remote_state.network.outputs.database_ip_filter
  }
}

[jylitalo]

https://github.com/UpCloudLtd/terraform-provider-upcloud/blob/1e5ad9aae496b21d1236c63bf8e84cc9420c46c6/CHANGELOG.md?plain=1#L90 lists pg_read_replica as removed, but if I do terraform state show upcloud_managed_database_postgresql.db | grep replica, it will only list:

        max_logical_replication_workers        = 0
        max_replication_slots                  = 0
            max_failover_replication_time_lag = 60

So nothing about pg_read_replica.

[jylitalo]

Also validated that terraform state pull | grep replica produces:

                "max_logical_replication_workers": 0,
                "max_replication_slots": 0,
                    "max_failover_replication_time_lag": 60
                "synchronous_replication": "",
                "allow_replication": false

[jylitalo]

Downgrade to 5.1.0 seems to resolve this issue.

[kangasta]

Hello, thanks for the bug report! Could you check if there is anything about pg_read_replica in your terraform.tfstate file(s)?

[jylitalo]

We are using postgresql as terraform backend, but terraform state pull should send us raw tfstate content. But I can still execute terraform plan/apply/refresh just fine with terraform provider 5.1.0 and everything breaks with 5.1.1 If I give terraform state show upcloud_managed_database_postgresql.db with 5.1.0, it looks like this:

# upcloud_managed_database_postgresql.db:
resource "upcloud_managed_database_postgresql" "db" {
    components              = [
        {
            component = "pg"
            host      = "REDACTED"
            port      = 11550
            route     = "dynamic"
            usage     = "primary"
        },
        {
            component = "pgbouncer"
            host      = "REDACTED"
            port      = 11551
            route     = "dynamic"
            usage     = "primary"
        },
    ]
    id                      = "096a683b-155d-4a62-9c01-bd3136424403"
    maintenance_window_dow  = "sunday"
    maintenance_window_time = "05:00:00"
    name                    = "REDACTED"
    node_states             = [
        {
            name  = "REDACTED"
            role  = "master"
            state = "running"
        },
    ]
    plan                    = "1x1xCPU-2GB-25GB"
    powered                 = true
    primary_database        = "defaultdb"
    service_host            = "REDACTED"
    service_password        = (sensitive value)
    service_port            = "11550"
    service_uri             = (sensitive value)
    service_username        = "worker"
    state                   = "running"
    title                   = "REDACTED"
    zone                    = "fi-hel2"

    properties {
        automatic_utility_network_ip_filter    = true
        autovacuum_analyze_scale_factor        = 0
        autovacuum_analyze_threshold           = 0
        autovacuum_freeze_max_age              = 0
        autovacuum_max_workers                 = 0
        autovacuum_naptime                     = 0
        autovacuum_vacuum_cost_delay           = 0
        autovacuum_vacuum_cost_limit           = 0
        autovacuum_vacuum_scale_factor         = 0
        autovacuum_vacuum_threshold            = 0
        backup_hour                            = 0
        backup_minute                          = 32
        bgwriter_delay                         = 0
        bgwriter_flush_after                   = 0
        bgwriter_lru_maxpages                  = 0
        bgwriter_lru_multiplier                = 0
        deadlock_timeout                       = 0
        idle_in_transaction_session_timeout    = 0
        ip_filter                              = [
            "REDACTED",
        ]
        jit                                    = false
        log_autovacuum_min_duration            = 0
        log_min_duration_statement             = 0
        log_temp_files                         = 0
        max_files_per_process                  = 0
        max_locks_per_transaction              = 0
        max_logical_replication_workers        = 0
        max_parallel_workers                   = 0
        max_parallel_workers_per_gather        = 0
        max_pred_locks_per_transaction         = 0
        max_prepared_transactions              = 0
        max_replication_slots                  = 0
        max_slot_wal_keep_size                 = 0
        max_stack_depth                        = 0
        max_standby_archive_delay              = 0
        max_standby_streaming_delay            = 0
        max_wal_senders                        = 0
        max_worker_processes                   = 0
        pg_partman_bgw_interval                = 0
        pg_stat_monitor_enable                 = false
        pg_stat_monitor_pgsm_enable_query_plan = false
        pg_stat_monitor_pgsm_max_buckets       = 0
        public_access                          = false
        service_log                            = false
        shared_buffers_percentage              = 0
        temp_file_limit                        = 0
        track_activity_query_size              = 0
        version                                = "15"
        wal_sender_timeout                     = 0
        wal_writer_delay                       = 0
        work_mem                               = 0

        pglookout {
            max_failover_replication_time_lag = 60
        }
    }
}

[villevsv-upcloud]

A patch release v5.2.1 for this bug has just been released.

We are happy to help out if any following issues occur.

[villevsv-upcloud]

Closing this for now, please come back if there are further issues.