mirrors/terraform-provider-binarylane
1
0
Fork 1
mirror of https://github.com/oscarhermoso/terraform-provider-binarylane synced 2026-10-21 22:09:15 +11:00
Code Packages Projects Releases Wiki Activity
Files
725fa71248646720854fee337ec38d2448208528
terraform-provider-binarylane/docs/data-sources/server.md
T
Oscar Hermoso 725fa71248 💽 Add disks attribute to server resource & data source 
Adds `disks` to `binarylane_server` and its data source, so a server's storage
can be split across several disks. Leave it unset and Binary Lane keeps managing
a single disk sized by `disk`; set it and it must describe every disk, including
the primary.

- The schema is generated the same way as `advanced_features`:
  scripts/data/server_disks.json is spliced into provider_code_spec.json by
  extend-provider.sh, with the list validator in
  internal/resources/server_disks_validators.go.
- `disk` remains the server's total storage. sum(disks[*].size_gigabytes) cannot
  exceed it, and any remainder is left unallocated.
- Create carves the additional disks out of the primary, which Binary Lane sizes
  to the whole total. Update frees space before the server's resize action and
  claims it afterwards, persisting what was actually done if a step fails.
- Disks are matched across plans by description, or by being the primary, and are
  numbered within a description so a server can have two disks sharing one.
- `disks` is returned in the order the configuration lists them, since Terraform
  requires a list's elements to stay where they were planned.
- Shrinking a disk warns at plan time, as does attaching additional disks
  alongside `user_data` that cloud-init may still be running.

Also bumps the OpenAPI schema to v0.39.2, and fixes two pre-existing bugs found
running the acceptance suite against real infrastructure: `advanced_features`
and `public_ipv6_addresses` weren't carried forward from state on a no-op
replan, so they planned as unknown on every apply.

🧪 Assert disks inline, and cover ordering in the acceptance tests

Drops the checkServerDisks/expectedDisk helper for the inline
TestCheckResourceAttr and TestCheckResourceAttrSet calls every other check in
this file uses.

Replaces the TestServerDisksToList unit test with the equivalent assertions in
TestServerResourceCreateWithDisks, which already reorders the disks of a real
server: the resource keeps the order its configuration lists them in, while the
data source, which has no order of its own, lists the primary disk first and the
rest by id. Both disks sharing the description "data" cover the duplicate case.

🧪 Pin disk ids across a list shift

Removing a disk shifts every disk after it down a position, which is where
terraform-plugin-framework#720 bites: Terraform merges prior state into the
proposed plan positionally, so the disk that moved up would inherit the id of
the one that was deleted, and the provider would resize the wrong disk.

planServerDisks matches by description (and by being the primary) instead of by
index, so this only asserts what should already hold: "alpha" and "beta" keep
the disks they had when "zeta" is removed from in front of them.

🏗️ Plan `disks` from a plan modifier on the attribute

Moves the planning of a configured `disks` out of the resource's ModifyPlan and
onto the attribute itself, as disksPlanModifier in internal/resources — the same
place, and declared the same way in scripts/data/server_disks.json, as
disksValidator.

Behaviour is unchanged: Terraform builds a list's proposed plan by merging prior
state in element by element, so removing a disk hands the disks after it the id
of their predecessor. The modifier plans the list from the configuration
instead, taking each disk's server-assigned values from the disk it matches in
prior state. DiskKey/DiskKeyer and DiskShrinkWarning move with it and are now
exported, since ordering and reconciliation match disks by the same rule.

What stays in ModifyPlan is the case `disks` is not configured, where what
happens to the disks depends on `disk` — the server's total, which ModifyPlan
resolves from state a few lines earlier, after attribute plan modifiers have
already run.

🏗️ Plan an unset `disks` on create from the plan modifier too

The creation case needs nothing but the absence of prior state, so it moves next
to the rest of the disk planning.

The remaining case — `disks` left out of the configuration on an existing server
— stays in ModifyPlan, and can't move: it depends on `disk`, the server's total,
which ModifyPlan resolves from state only after attribute plan modifiers have
run. Reading `disk` from the plan inside the modifier plans the disks as unknown
on every no-op replan, which TestServerResourceRename catches:

    ~ disks = [ - { id = 1 ... } ] -> (known after apply)
    Plan: 0 to add, 1 to change, 0 to destroy.

Working the total out from the configuration inside the modifier would restate a
rule that already lives in ModifyPlan, so it stays where the value it needs is.

🧪 Try `disks` as a SetNestedAttribute

A set has no order, so everything that existed to keep a list's elements where
Terraform planned them goes away: serverDisksToList's ranking and its `order`
argument threaded through five call sites become a three-line
types.SetValueFrom, and a configuration can list its disks however it likes.

Verified against the live API: create with duplicate descriptions, update
(resize, add, remove), dropping `disks`, reordering, and resizing one of two
disks that share a description — the case a set is least equipped for, since it
has no positions to match duplicates by.

Costs: a resize renders as a remove/add pair rather than an in-place change (the
id is preserved in both halves), and tests need checkDiskId to find a disk,
since a set has no index to address one with.

🪞 Reconcile disks as binarylane.Disk, not as Terraform values

binarylane.Disk carries tfsdk tags for exactly the attributes `disks` has, so a
configured set reflects straight into it. splitDisks now hands Create and Update
the API's own type, which is what the disks they compare against already are:
the comparison that read `disk.SizeGigabytes.ValueInt64() <
int64(other.SizeGigabytes)` now reads `disk.SizeGigabytes <
other.SizeGigabytes`. Eight value-unwrapping conversions go with it.

The plan modifier keeps building Terraform values by hand, since the attributes
Binary Lane assigns are still unknown while it runs.

🧪 Restore `disks` to ListNestedAttribute

Co-Authored-By: Marais Rossouw <maraisr@users.noreply.github.com>
2026-08-30 18:12:02 +08:00

5.6 KiB
Raw Blame History

page_title, subcategory, description
page_title subcategory description
binarylane_server Data Source - terraform-provider-binarylane Retrieve details about a BinaryLane Server.

binarylane_server (Data Source)

Retrieve details about a BinaryLane Server.

Example Usage

data "binarylane_server" "example" {
  id = 123456
}

Schema

Required

  • id (Number) The ID of the server to fetch.

Read-Only

  • advanced_features (Object) (see below for nested schema)
  • backups (Boolean) If true this will enable two daily backups for the server. By default, backups are disabled.
  • disk (Number) The total storage in GB for this server. Leave null to accept the default for the size. Valid values:
    • must be a multiple of 5
    • > 60 GB must be a multiple of 10
    • > 200 GB must be a multiple of 100
  • disks (Attributes List) The disks attached to this server. Leave null to let Binary Lane manage the server's disks. If specified, every disk must be listed, including the primary: exactly one entry must have primary = true, and its id and description are assigned by Binary Lane. (see below for nested schema)
  • image (String) The slug of the selected operating system, such as debian-12. You can fetch a full list of images from the BinaryLane API.
  • ipv6 (Boolean) If true this will add a public and private IPv6 address to the server. By default, IPv6 is disabled.
  • memory (Number) The total memory in MB for this server. Leave null to accept the default size. Valid values:
    • must be a multiple of 128
    • > 2048 MB must be a multiple of 1024
    • > 16384 MB must be a multiple of 2048
    • > 24576 MB must be a multiple of 4096
  • name (String) The hostname of your server, such as vps01.yourcompany.com. If not provided, the server will be created with a random name.
  • permalink (String) A randomly generated two-word identifier assigned to servers in regions that support this feature
  • port_blocking (Boolean) Port blocking of outgoing connections for email, SSH and Remote Desktop (TCP ports 22, 25, and 3389) is enabled by default for all new servers. If this is false port blocking will be disabled. Disabling port blocking is only available to verified accounts.
  • private_ipv4_addresses (List of String) The private IPv4 addresses assigned to the server.
  • private_ipv6_addresses (List of String) The private IPv6 addresses assigned to the server.
  • public_ipv4_addresses (List of String) The public IPv4 addresses assigned to the server.
  • public_ipv6_addresses (List of String) The public IPv6 addresses assigned to the server.
  • region (String) The slug of the selected region.
  • separate_private_network_interface (Boolean) This attribute can only be set if your server also has a vpc_id attribute set. When enabled, a separate private network interface is provided for the server's VPC traffic.
  • size (String) The slug of the selected size.
  • source_and_destination_check (Boolean) This attribute can only be set if your server also has a vpc_id attribute set. When enabled (which is true by default), your server will only be able to send or receive packets that are directly addressed to one of the IP addresses associated with the Cloud Server. Generally, this is desirable behaviour because it prevents IP conflicts and other hard-to-diagnose networking faults due to incorrect network configuration. When source_and_destination_check is false, your Cloud Server will be able to send and receive packets addressed to any server. This is typically used when you want to use your Cloud Server as a VPN endpoint, a NAT server to provide internet access, or IP forwarding.
  • ssh_keys (List of Number) This is a list of SSH key ids. If this is null or not provided, any SSH keys that have been marked as default will be deployed (assuming the operating system supports SSH Keys). Submit an empty list to disable deployment of default keys.
  • user_data (String) A script or cloud-config YAML file to configure the server. Can only be specified if the OS image supports UserData (i.e. not Windows). See more: https://cloudinit.readthedocs.io/en/latest/explanation/format.html#user-data-script
  • vpc_id (Number) Leave null to use default (public) network for the selected region.
  • vpc_ipv4_address (String) If provided this will be the IPv4 address for the server's private VPC network adapter. If this is unspecified, then an unused IPv4 address will be assigned. This field is only valid when vpc_id is provided.

Nested Schema for advanced_features

Read-Only:

  • cloud_init (Boolean)
  • driver_disk (Boolean)
  • emulated_devices (Boolean)
  • emulated_hyperv (Boolean)
  • emulated_tpm (Boolean)
  • local_rtc (Boolean)
  • nested_virt (Boolean)
  • qemu_guest_agent (Boolean)
  • uefi_boot (Boolean)
  • unset_uuid (Boolean)

Nested Schema for disks

Read-Only:

  • description (String) A label for this disk. Assigned by Binary Lane for the primary disk, so must be left null when primary = true. Changing this will destroy the disk and create a new one, losing any data on it. Disks that share a description are matched in the order they are listed.
  • id (Number) The ID of this disk.
  • primary (Boolean) If true this is the primary disk, which the operating system is installed on. Exactly one disk must be the primary.
  • size_gigabytes (Number) The size of the disk in GB. The sum across all disks cannot exceed disk, and any remainder is left unallocated.
View Git Blame Copy Permalink