terraform/states/state.go

230 lines
6.8 KiB
Go
Raw Normal View History

states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
package states
import (
various: helpers for collecting necessary provider types Since schemas are required to interpret provider, resource, and provisioner attributes in configs, states, and plans, these helpers intend to make it easier to gather up the the necessary provider types in order to preload all of the needed schemas before beginning further processing. Config.ProviderTypes returns directly the list of provider types, since at this level further detail is not useful: we've not yet run the provider allocation algorithm, and so the only thing we can reliably extract here is provider types themselves. State.ProviderAddrs and Plan.ProviderAddrs each return a list of absolute provider addresses, which can then be turned into a list of provider types using the new helper providers.AddressedTypesAbs. Since we're already using configs.Config throughout core, this also updates the terraform.LoadSchemas helper to use Config.ProviderTypes to find the necessary providers, rather than implementing its own discovery logic. states.State is not yet plumbed in, so we cannot yet use State.ProviderAddrs to deal with the state but there's a TODO comment to remind us to update that in a later commit when we swap out terraform.State for states.State. A later commit will probably refactor this further so that we can easily obtain schema for the providers needed to interpret a plan too, but that is deferred here because further work is required to make core work with the new plan types first. At that point, terraform.LoadSchemas may become providers.LoadSchemas with a different interface that just accepts lists of provider and provisioner names that have been gathered by the caller using these new helpers.
2018-06-22 02:39:27 +02:00
"sort"
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
"github.com/zclconf/go-cty/cty"
"github.com/hashicorp/terraform/addrs"
)
// State is the top-level type of a Terraform state.
//
// A state should be mutated only via its accessor methods, to ensure that
// invariants are preserved.
//
// Access to State and the nested values within it is not concurrency-safe,
// so when accessing a State object concurrently it is the caller's
// responsibility to ensure that only one write is in progress at a time
// and that reads only occur when no write is in progress. The most common
// way to acheive this is to wrap the State in a SyncState and use the
// higher-level atomic operations supported by that type.
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
type State struct {
// Modules contains the state for each module. The keys in this map are
// an implementation detail and must not be used by outside callers.
Modules map[string]*Module
}
// NewState constructs a minimal empty state, containing an empty root module.
func NewState() *State {
modules := map[string]*Module{}
modules[addrs.RootModuleInstance.String()] = NewModule(addrs.RootModuleInstance)
return &State{
Modules: modules,
}
}
// BuildState is a helper -- primarily intended for tests -- to build a state
// using imperative code against the StateSync type while still acting as
// an expression of type *State to assign into a containing struct.
func BuildState(cb func(*SyncState)) *State {
s := NewState()
cb(s.SyncWrapper())
return s
}
// Empty returns true if there are no resources or populated output values
// in the receiver. In other words, if this state could be safely replaced
// with the return value of NewState and be functionally equivalent.
func (s *State) Empty() bool {
if s == nil {
return true
}
for _, ms := range s.Modules {
if len(ms.Resources) != 0 {
return false
}
if len(ms.OutputValues) != 0 {
return false
}
}
return true
}
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
// Module returns the state for the module with the given address, or nil if
// the requested module is not tracked in the state.
func (s *State) Module(addr addrs.ModuleInstance) *Module {
core: Be more explicit in how we handle create_before_destroy Previously our handling of create_before_destroy -- and of deposed objects in particular -- was rather "implicit" and spread over various different subsystems. We'd quietly just destroy every deposed object during a destroy operation, without any user-visible plan to do so. Here we make things more explicit by tracking each deposed object individually by its pseudorandomly-allocated key. There are two different mechanisms at play here, building on the same concepts: - During a replace operation with create_before_destroy, we *pre-allocate* a DeposedKey to use for the prior object in the "apply" node and then pass that exact id to the destroy node, ensuring that we only destroy the single object we planned to destroy. In the happy path here the user never actually sees the allocated deposed key because we use it and then immediately destroy it within the same operation. However, that destroy may fail, which brings us to the second mechanism: - If any deposed objects are already present in state during _plan_, we insert a destroy change for them into the plan so that it's explicit to the user that we are going to destroy these additional objects, and then create an individual graph node for each one in DiffTransformer. The main motivation here is to be more careful in how we handle these destroys so that from a user's standpoint we never destroy something without the user knowing about it ahead of time. However, this new organization also hopefully makes the code itself a little easier to follow because the connection between the create and destroy steps of a Replace is reprseented in a single place (in DiffTransformer) and deposed instances each have their own explicit graph node rather than being secretly handled as part of the main instance-level graph node.
2018-09-20 21:30:52 +02:00
if s == nil {
panic("State.Module on nil *State")
}
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
return s.Modules[addr.String()]
}
// RemoveModule removes the module with the given address from the state,
// unless it is the root module. The root module cannot be deleted, and so
// this method will panic if that is attempted.
//
// Removing a module implicitly discards all of the resources, outputs and
// local values within it, and so this should usually be done only for empty
// modules. For callers accessing the state through a SyncState wrapper, modules
// are automatically pruned if they are empty after one of their contained
// elements is removed.
func (s *State) RemoveModule(addr addrs.ModuleInstance) {
if addr.IsRoot() {
panic("attempted to remove root module")
}
delete(s.Modules, addr.String())
}
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
// RootModule is a convenient alias for Module(addrs.RootModuleInstance).
func (s *State) RootModule() *Module {
if s == nil {
panic("RootModule called on nil State")
}
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
return s.Modules[addrs.RootModuleInstance.String()]
}
// EnsureModule returns the state for the module with the given address,
// creating and adding a new one if necessary.
//
// Since this might modify the state to add a new instance, it is considered
// to be a write operation.
func (s *State) EnsureModule(addr addrs.ModuleInstance) *Module {
ms := s.Module(addr)
if ms == nil {
ms = NewModule(addr)
s.Modules[addr.String()] = ms
}
return ms
}
// HasResources returns true if there is at least one resource (of any mode)
// present in the receiving state.
func (s *State) HasResources() bool {
if s == nil {
return false
}
for _, ms := range s.Modules {
if len(ms.Resources) > 0 {
return true
}
}
return false
}
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
// Resource returns the state for the resource with the given address, or nil
// if no such resource is tracked in the state.
func (s *State) Resource(addr addrs.AbsResource) *Resource {
ms := s.Module(addr.Module)
if ms == nil {
return nil
}
return ms.Resource(addr.Resource)
}
// ResourceInstance returns the state for the resource instance with the given
// address, or nil if no such resource is tracked in the state.
func (s *State) ResourceInstance(addr addrs.AbsResourceInstance) *ResourceInstance {
core: Be more explicit in how we handle create_before_destroy Previously our handling of create_before_destroy -- and of deposed objects in particular -- was rather "implicit" and spread over various different subsystems. We'd quietly just destroy every deposed object during a destroy operation, without any user-visible plan to do so. Here we make things more explicit by tracking each deposed object individually by its pseudorandomly-allocated key. There are two different mechanisms at play here, building on the same concepts: - During a replace operation with create_before_destroy, we *pre-allocate* a DeposedKey to use for the prior object in the "apply" node and then pass that exact id to the destroy node, ensuring that we only destroy the single object we planned to destroy. In the happy path here the user never actually sees the allocated deposed key because we use it and then immediately destroy it within the same operation. However, that destroy may fail, which brings us to the second mechanism: - If any deposed objects are already present in state during _plan_, we insert a destroy change for them into the plan so that it's explicit to the user that we are going to destroy these additional objects, and then create an individual graph node for each one in DiffTransformer. The main motivation here is to be more careful in how we handle these destroys so that from a user's standpoint we never destroy something without the user knowing about it ahead of time. However, this new organization also hopefully makes the code itself a little easier to follow because the connection between the create and destroy steps of a Replace is reprseented in a single place (in DiffTransformer) and deposed instances each have their own explicit graph node rather than being secretly handled as part of the main instance-level graph node.
2018-09-20 21:30:52 +02:00
if s == nil {
panic("State.ResourceInstance on nil *State")
}
states: New package with modern models for Terraform state Our previous state models in the "terraform" package had a few limitations that are addressed here: - Instance attributes were stored as map[string]string with dot-separated keys representing traversals through a data structure. Now that we have a full type system, it's preferable to store it as a real data structure. - The existing state structures skipped over the "resource" concept and went straight to resource instance, requiring heuristics to decide whether a particular resource should appear as a single object or as a list of objects when used in configuration expressions. - Related to the previous point, the state models also used incorrect terminology where "ResourceState" was really a resource instance state and "InstanceState" was really the state of a particular remote object associated with an instance. These new models use the correct names for each of these, introducing the idea of a "ResourceInstanceObject" as the local record of a remote object associated with an instance. This is a first pass at fleshing out a new model for state. Undoubtedly there will be further iterations of this as we work on integrating these new models into the "terraform" package. These new model types no longer serve double-duty as a description of the JSON state file format, since they are for in-memory use only. A subsequent commit will introduce a separate package that deals with persisting state to files and reloading those files later.
2018-06-08 02:27:57 +02:00
ms := s.Module(addr.Module)
if ms == nil {
return nil
}
return ms.ResourceInstance(addr.Resource)
}
// OutputValue returns the state for the output value with the given address,
// or nil if no such output value is tracked in the state.
func (s *State) OutputValue(addr addrs.AbsOutputValue) *OutputValue {
ms := s.Module(addr.Module)
if ms == nil {
return nil
}
return ms.OutputValues[addr.OutputValue.Name]
}
// LocalValue returns the value of the named local value with the given address,
// or cty.NilVal if no such value is tracked in the state.
func (s *State) LocalValue(addr addrs.AbsLocalValue) cty.Value {
ms := s.Module(addr.Module)
if ms == nil {
return cty.NilVal
}
return ms.LocalValues[addr.LocalValue.Name]
}
various: helpers for collecting necessary provider types Since schemas are required to interpret provider, resource, and provisioner attributes in configs, states, and plans, these helpers intend to make it easier to gather up the the necessary provider types in order to preload all of the needed schemas before beginning further processing. Config.ProviderTypes returns directly the list of provider types, since at this level further detail is not useful: we've not yet run the provider allocation algorithm, and so the only thing we can reliably extract here is provider types themselves. State.ProviderAddrs and Plan.ProviderAddrs each return a list of absolute provider addresses, which can then be turned into a list of provider types using the new helper providers.AddressedTypesAbs. Since we're already using configs.Config throughout core, this also updates the terraform.LoadSchemas helper to use Config.ProviderTypes to find the necessary providers, rather than implementing its own discovery logic. states.State is not yet plumbed in, so we cannot yet use State.ProviderAddrs to deal with the state but there's a TODO comment to remind us to update that in a later commit when we swap out terraform.State for states.State. A later commit will probably refactor this further so that we can easily obtain schema for the providers needed to interpret a plan too, but that is deferred here because further work is required to make core work with the new plan types first. At that point, terraform.LoadSchemas may become providers.LoadSchemas with a different interface that just accepts lists of provider and provisioner names that have been gathered by the caller using these new helpers.
2018-06-22 02:39:27 +02:00
// ProviderAddrs returns a list of all of the provider configuration addresses
// referenced throughout the receiving state.
//
// The result is de-duplicated so that each distinct address appears only once.
func (s *State) ProviderAddrs() []addrs.AbsProviderConfig {
if s == nil {
return nil
}
m := map[string]addrs.AbsProviderConfig{}
for _, ms := range s.Modules {
for _, rc := range ms.Resources {
m[rc.ProviderConfig.String()] = rc.ProviderConfig
}
}
if len(m) == 0 {
return nil
}
// This is mainly just so we'll get stable results for testing purposes.
keys := make([]string, 0, len(m))
for k := range m {
keys = append(keys, k)
}
sort.Strings(keys)
ret := make([]addrs.AbsProviderConfig, len(keys))
for i, key := range keys {
ret[i] = m[key]
}
return ret
}
// PruneResourceHusks is a specialized method that will remove any Resource
// objects that do not contain any instances, even if they have an EachMode.
//
// This should generally be used only after a "terraform destroy" operation,
// to finalize the cleanup of the state. It is not correct to use this after
// other operations because if a resource has "count = 0" or "for_each" over
// an empty collection then we want to retain it in the state so that references
// to it, particularly in "strange" contexts like "terraform console", can be
// properly resolved.
//
// This method MUST NOT be called concurrently with other readers and writers
// of the receiving state.
func (s *State) PruneResourceHusks() {
for _, m := range s.Modules {
m.PruneResourceHusks()
if len(m.Resources) == 0 && !m.Addr.IsRoot() {
s.RemoveModule(m.Addr)
}
}
}
// SyncWrapper returns a SyncState object wrapping the receiver.
func (s *State) SyncWrapper() *SyncState {
return &SyncState{
state: s,
}
}