statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
package statefile
import (
"encoding/json"
"fmt"
"strconv"
"strings"
2019-11-21 01:53:56 +01:00
"github.com/hashicorp/hcl/v2/hclsyntax"
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
"github.com/zclconf/go-cty/cty"
ctyjson "github.com/zclconf/go-cty/cty/json"
2021-05-17 21:00:50 +02:00
"github.com/hashicorp/terraform/internal/addrs"
2021-05-17 21:17:09 +02:00
"github.com/hashicorp/terraform/internal/configs"
2021-05-17 19:11:06 +02:00
"github.com/hashicorp/terraform/internal/tfdiags"
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
"github.com/hashicorp/terraform/states"
)
func upgradeStateV3ToV4 ( old * stateV3 ) ( * stateV4 , error ) {
if old . Serial < 0 {
// The new format is using uint64 here, which should be fine for any
// real state (we only used positive integers in practice) but we'll
// catch this explicitly here to avoid weird behavior if a state file
// has been tampered with in some way.
return nil , fmt . Errorf ( "state has serial less than zero, which is invalid" )
}
new := & stateV4 {
TerraformVersion : old . TFVersion ,
Serial : uint64 ( old . Serial ) ,
Lineage : old . Lineage ,
RootOutputs : map [ string ] outputStateV4 { } ,
Resources : [ ] resourceStateV4 { } ,
}
if new . TerraformVersion == "" {
// Older formats considered this to be optional, but now it's required
// and so we'll stub it out with something that's definitely older
// than the version that really created this state.
new . TerraformVersion = "0.0.0"
}
for _ , msOld := range old . Modules {
if len ( msOld . Path ) < 1 || msOld . Path [ 0 ] != "root" {
return nil , fmt . Errorf ( "state contains invalid module path %#v" , msOld . Path )
}
// Convert legacy-style module address into our newer address type.
// Since these old formats are only generated by versions of Terraform
// that don't support count and for_each on modules, we can just assume
// all of the modules are unkeyed.
moduleAddr := make ( addrs . ModuleInstance , len ( msOld . Path ) - 1 )
for i , name := range msOld . Path [ 1 : ] {
2019-11-21 01:53:56 +01:00
if ! hclsyntax . ValidIdentifier ( name ) {
// If we don't fail here then we'll produce an invalid state
// version 4 which subsequent operations will reject, so we'll
// fail early here for safety to make sure we can never
// inadvertently commit an invalid snapshot to a backend.
return nil , fmt . Errorf ( "state contains invalid module path %#v: %q is not a valid identifier; rename it in Terraform 0.11 before upgrading to Terraform 0.12" , msOld . Path , name )
}
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
moduleAddr [ i ] = addrs . ModuleInstanceStep {
Name : name ,
InstanceKey : addrs . NoKey ,
}
}
// In a v3 state file, a "resource state" is actually an instance
2021-02-05 15:11:38 +01:00
// state, so we need to fill in a missing level of hierarchy here
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
// by lazily creating resource states as we encounter them.
// We'll track them in here, keyed on the string representation of
// the resource address.
resourceStates := map [ string ] * resourceStateV4 { }
for legacyAddr , rsOld := range msOld . Resources {
instAddr , err := parseLegacyResourceAddress ( legacyAddr )
if err != nil {
return nil , err
}
resAddr := instAddr . Resource
rs , exists := resourceStates [ resAddr . String ( ) ]
if ! exists {
var modeStr string
switch resAddr . Mode {
case addrs . ManagedResourceMode :
modeStr = "managed"
case addrs . DataResourceMode :
modeStr = "data"
default :
2019-07-19 02:35:47 +02:00
return nil , fmt . Errorf ( "state contains resource %s with an unsupported resource mode %#v" , resAddr , resAddr . Mode )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
// In state versions prior to 4 we allowed each instance of a
// resource to have its own provider configuration address,
// which makes no real sense in practice because providers
// are associated with resources in the configuration. We
// elevate that to the resource level during this upgrade,
// implicitly taking the provider address of the first instance
// we encounter for each resource. While this is lossy in
// theory, in practice there is no reason for these values to
// differ between instances.
var providerAddr addrs . AbsProviderConfig
oldProviderAddr := rsOld . Provider
if strings . Contains ( oldProviderAddr , "provider." ) {
// Smells like a new-style provider address, but we'll test it.
var diags tfdiags . Diagnostics
2020-02-13 21:32:58 +01:00
providerAddr , diags = addrs . ParseLegacyAbsProviderConfigStr ( oldProviderAddr )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
if diags . HasErrors ( ) {
2019-11-21 01:53:56 +01:00
if strings . Contains ( oldProviderAddr , "${" ) {
// There seems to be a common misconception that
// interpolation was valid in provider aliases
// in 0.11, so we'll use a specialized error
// message for that case.
return nil , fmt . Errorf ( "invalid provider config reference %q for %s: this alias seems to contain a template interpolation sequence, which was not supported but also not error-checked in Terraform 0.11. To proceed, rename the associated provider alias to a valid identifier and apply the change with Terraform 0.11 before upgrading to Terraform 0.12" , oldProviderAddr , instAddr )
}
2019-07-19 02:35:47 +02:00
return nil , fmt . Errorf ( "invalid provider config reference %q for %s: %s" , oldProviderAddr , instAddr , diags . Err ( ) )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
} else {
// Smells like an old-style module-local provider address,
// which we'll need to migrate. We'll assume it's referring
// to the same module the resource is in, which might be
// incorrect but it'll get fixed up next time any updates
// are made to an instance.
if oldProviderAddr != "" {
2019-12-06 18:14:54 +01:00
localAddr , diags := configs . ParseProviderConfigCompactStr ( oldProviderAddr )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
if diags . HasErrors ( ) {
2019-11-21 01:53:56 +01:00
if strings . Contains ( oldProviderAddr , "${" ) {
// There seems to be a common misconception that
// interpolation was valid in provider aliases
// in 0.11, so we'll use a specialized error
// message for that case.
return nil , fmt . Errorf ( "invalid legacy provider config reference %q for %s: this alias seems to contain a template interpolation sequence, which was not supported but also not error-checked in Terraform 0.11. To proceed, rename the associated provider alias to a valid identifier and apply the change with Terraform 0.11 before upgrading to Terraform 0.12" , oldProviderAddr , instAddr )
}
2019-07-19 02:35:47 +02:00
return nil , fmt . Errorf ( "invalid legacy provider config reference %q for %s: %s" , oldProviderAddr , instAddr , diags . Err ( ) )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
2020-02-13 21:32:58 +01:00
providerAddr = addrs . AbsProviderConfig {
2020-04-03 20:20:24 +02:00
Module : moduleAddr . Module ( ) ,
// We use NewLegacyProvider here so we can use
// LegacyString() below to get the appropriate
// legacy-style provider string.
2020-02-13 21:32:58 +01:00
Provider : addrs . NewLegacyProvider ( localAddr . LocalName ) ,
Alias : localAddr . Alias ,
}
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
} else {
Initial steps towards AbsProviderConfig/LocalProviderConfig separation (#23978)
* Introduce "Local" terminology for non-absolute provider config addresses
In a future change AbsProviderConfig and LocalProviderConfig are going to
become two entirely distinct types, rather than Abs embedding Local as
written here. This naming change is in preparation for that subsequent
work, which will also include introducing a new "ProviderConfig" type
that is an interface that AbsProviderConfig and LocalProviderConfig both
implement.
This is intended to be largely just a naming change to get started, so
we can deal with all of the messy renaming. However, this did also require
a slight change in modeling where the Resource.DefaultProviderConfig
method has become Resource.DefaultProvider returning a Provider address
directly, because this method doesn't have enough information to construct
a true and accurate LocalProviderConfig -- it would need to refer to the
configuration to know what this module is calling the provider it has
selected.
In order to leave a trail to follow for subsequent work, all of the
changes here are intended to ensure that remaining work will become
obvious via compile-time errors when all of the following changes happen:
- The concept of "legacy" provider addresses is removed from the addrs
package, including removing addrs.NewLegacyProvider and
addrs.Provider.LegacyString.
- addrs.AbsProviderConfig stops having addrs.LocalProviderConfig embedded
in it and has an addrs.Provider and a string alias directly instead.
- The provider-schema-handling parts of Terraform core are updated to
work with addrs.Provider to identify providers, rather than legacy
strings.
In particular, there are still several codepaths here making legacy
provider address assumptions (in order to limit the scope of this change)
but I've made sure each one is doing something that relies on at least
one of the above changes not having been made yet.
* addrs: ProviderConfig interface
In a (very) few special situations in the main "terraform" package we need
to make runtime decisions about whether a provider config is absolute
or local.
We currently do that by exploiting the fact that AbsProviderConfig has
LocalProviderConfig nested inside of it and so in the local case we can
just ignore the wrapping AbsProviderConfig and use the embedded value.
In a future change we'll be moving away from that embedding and making
these two types distinct in order to represent that mapping between them
requires consulting a lookup table in the configuration, and so here we
introduce a new interface type ProviderConfig that can represent either
AbsProviderConfig or LocalProviderConfig decided dynamically at runtime.
This also includes the Config.ResolveAbsProviderAddr method that will
eventually be responsible for that local-to-absolute translation, so
that callers with access to the configuration can normalize to an
addrs.AbsProviderConfig given a non-nil addrs.ProviderConfig. That's
currently unused because existing callers are still relying on the
simplistic structural transform, but we'll switch them over in a later
commit.
* rename LocalType to LocalName
Co-authored-by: Kristin Laemmert <mildwonkey@users.noreply.github.com>
2020-01-31 14:23:07 +01:00
providerAddr = addrs . AbsProviderConfig {
2020-04-03 20:20:24 +02:00
Module : moduleAddr . Module ( ) ,
// We use NewLegacyProvider here so we can use
// LegacyString() below to get the appropriate
// legacy-style provider string.
2020-03-18 13:58:20 +01:00
Provider : addrs . NewLegacyProvider ( resAddr . ImpliedProvider ( ) ) ,
Initial steps towards AbsProviderConfig/LocalProviderConfig separation (#23978)
* Introduce "Local" terminology for non-absolute provider config addresses
In a future change AbsProviderConfig and LocalProviderConfig are going to
become two entirely distinct types, rather than Abs embedding Local as
written here. This naming change is in preparation for that subsequent
work, which will also include introducing a new "ProviderConfig" type
that is an interface that AbsProviderConfig and LocalProviderConfig both
implement.
This is intended to be largely just a naming change to get started, so
we can deal with all of the messy renaming. However, this did also require
a slight change in modeling where the Resource.DefaultProviderConfig
method has become Resource.DefaultProvider returning a Provider address
directly, because this method doesn't have enough information to construct
a true and accurate LocalProviderConfig -- it would need to refer to the
configuration to know what this module is calling the provider it has
selected.
In order to leave a trail to follow for subsequent work, all of the
changes here are intended to ensure that remaining work will become
obvious via compile-time errors when all of the following changes happen:
- The concept of "legacy" provider addresses is removed from the addrs
package, including removing addrs.NewLegacyProvider and
addrs.Provider.LegacyString.
- addrs.AbsProviderConfig stops having addrs.LocalProviderConfig embedded
in it and has an addrs.Provider and a string alias directly instead.
- The provider-schema-handling parts of Terraform core are updated to
work with addrs.Provider to identify providers, rather than legacy
strings.
In particular, there are still several codepaths here making legacy
provider address assumptions (in order to limit the scope of this change)
but I've made sure each one is doing something that relies on at least
one of the above changes not having been made yet.
* addrs: ProviderConfig interface
In a (very) few special situations in the main "terraform" package we need
to make runtime decisions about whether a provider config is absolute
or local.
We currently do that by exploiting the fact that AbsProviderConfig has
LocalProviderConfig nested inside of it and so in the local case we can
just ignore the wrapping AbsProviderConfig and use the embedded value.
In a future change we'll be moving away from that embedding and making
these two types distinct in order to represent that mapping between them
requires consulting a lookup table in the configuration, and so here we
introduce a new interface type ProviderConfig that can represent either
AbsProviderConfig or LocalProviderConfig decided dynamically at runtime.
This also includes the Config.ResolveAbsProviderAddr method that will
eventually be responsible for that local-to-absolute translation, so
that callers with access to the configuration can normalize to an
addrs.AbsProviderConfig given a non-nil addrs.ProviderConfig. That's
currently unused because existing callers are still relying on the
simplistic structural transform, but we'll switch them over in a later
commit.
* rename LocalType to LocalName
Co-authored-by: Kristin Laemmert <mildwonkey@users.noreply.github.com>
2020-01-31 14:23:07 +01:00
}
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
}
rs = & resourceStateV4 {
Module : moduleAddr . String ( ) ,
Mode : modeStr ,
Type : resAddr . Type ,
Name : resAddr . Name ,
Instances : [ ] instanceObjectStateV4 { } ,
2020-03-31 14:33:48 +02:00
ProviderConfig : providerAddr . LegacyString ( ) ,
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
resourceStates [ resAddr . String ( ) ] = rs
}
// Now we'll deal with the instance itself, which may either be
// the first instance in a resource we just created or an additional
// instance for a resource added on a prior loop.
instKey := instAddr . Key
if isOld := rsOld . Primary ; isOld != nil {
isNew , err := upgradeInstanceObjectV3ToV4 ( rsOld , isOld , instKey , states . NotDeposed )
if err != nil {
return nil , fmt . Errorf ( "failed to migrate primary generation of %s: %s" , instAddr , err )
}
rs . Instances = append ( rs . Instances , * isNew )
}
for i , isOld := range rsOld . Deposed {
// When we migrate old instances we'll use sequential deposed
// keys just so that the upgrade result is deterministic. New
// deposed keys allocated moving forward will be pseudorandomly
// selected, but we check for collisions and so these
// non-random ones won't hurt.
deposedKey := states . DeposedKey ( fmt . Sprintf ( "%08x" , i + 1 ) )
isNew , err := upgradeInstanceObjectV3ToV4 ( rsOld , isOld , instKey , deposedKey )
if err != nil {
return nil , fmt . Errorf ( "failed to migrate deposed generation index %d of %s: %s" , i , instAddr , err )
}
rs . Instances = append ( rs . Instances , * isNew )
}
if instKey != addrs . NoKey && rs . EachMode == "" {
rs . EachMode = "list"
}
}
for _ , rs := range resourceStates {
new . Resources = append ( new . Resources , * rs )
}
if len ( msOld . Path ) == 1 && msOld . Path [ 0 ] == "root" {
// We'll migrate the outputs for this module too, then.
for name , oldOS := range msOld . Outputs {
newOS := outputStateV4 {
Sensitive : oldOS . Sensitive ,
}
valRaw := oldOS . Value
valSrc , err := json . Marshal ( valRaw )
if err != nil {
// Should never happen, because this value came from JSON
// in the first place and so we're just round-tripping here.
return nil , fmt . Errorf ( "failed to serialize output %q value as JSON: %s" , name , err )
}
// The "type" field in state V2 wasn't really that useful
// since it was only able to capture string vs. list vs. map.
// For this reason, during upgrade we'll just discard it
// altogether and use cty's idea of the implied type of
// turning our old value into JSON.
ty , err := ctyjson . ImpliedType ( valSrc )
if err != nil {
// REALLY should never happen, because we literally just
// encoded this as JSON above!
return nil , fmt . Errorf ( "failed to parse output %q value from JSON: %s" , name , err )
}
// ImpliedType tends to produce structural types, but since older
// version of Terraform didn't support those a collection type
// is probably what was intended, so we'll see if we can
// interpret our value as one.
ty = simplifyImpliedValueType ( ty )
tySrc , err := ctyjson . MarshalType ( ty )
if err != nil {
return nil , fmt . Errorf ( "failed to serialize output %q type as JSON: %s" , name , err )
}
newOS . ValueRaw = json . RawMessage ( valSrc )
newOS . ValueTypeRaw = json . RawMessage ( tySrc )
new . RootOutputs [ name ] = newOS
}
}
}
new . normalize ( )
return new , nil
}
func upgradeInstanceObjectV3ToV4 ( rsOld * resourceStateV2 , isOld * instanceStateV2 , instKey addrs . InstanceKey , deposedKey states . DeposedKey ) ( * instanceObjectStateV4 , error ) {
// Schema versions were, in prior formats, a private concern of the provider
// SDK, and not a first-class concept in the state format. Here we're
// sniffing for the pre-0.12 SDK's way of representing schema versions
// and promoting it to our first-class field if we find it. We'll ignore
terraform: Ugly huge change to weave in new State and Plan types
Due to how often the state and plan types are referenced throughout
Terraform, there isn't a great way to switch them out gradually. As a
consequence, this huge commit gets us from the old world to a _compilable_
new world, but still has a large number of known test failures due to
key functionality being stubbed out.
The stubs here are for anything that interacts with providers, since we
now need to do the follow-up work to similarly replace the old
terraform.ResourceProvider interface with its replacement in the new
"providers" package. That work, along with work to fix the remaining
failing tests, will follow in subsequent commits.
The aim here was to replace all references to terraform.State and its
downstream types with states.State, terraform.Plan with plans.Plan,
state.State with statemgr.State, and switch to the new implementations of
the state and plan file formats. However, due to the number of times those
types are used, this also ended up affecting numerous other parts of core
such as terraform.Hook, the backend.Backend interface, and most of the CLI
commands.
Just as with 5861dbf3fc49b19587a31816eb06f511ab861bb4 before, I apologize
in advance to the person who inevitably just found this huge commit while
spelunking through the commit history.
2018-08-14 23:24:45 +02:00
// it if it doesn't look like what the SDK would've written. If this
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
// sniffing fails then we'll assume schema version 0.
var schemaVersion uint64
migratedSchemaVersion := false
if raw , exists := isOld . Meta [ "schema_version" ] ; exists {
2018-08-18 01:11:07 +02:00
switch tv := raw . ( type ) {
case string :
v , err := strconv . ParseUint ( tv , 10 , 64 )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
if err == nil {
schemaVersion = v
migratedSchemaVersion = true
}
2018-08-18 01:11:07 +02:00
case int :
schemaVersion = uint64 ( tv )
migratedSchemaVersion = true
case float64 :
schemaVersion = uint64 ( tv )
migratedSchemaVersion = true
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
}
private := map [ string ] interface { } { }
for k , v := range isOld . Meta {
if k == "schema_version" && migratedSchemaVersion {
// We're gonna promote this into our first-class schema version field
continue
}
private [ k ] = v
}
var privateJSON [ ] byte
if len ( private ) != 0 {
var err error
privateJSON , err = json . Marshal ( private )
if err != nil {
// This shouldn't happen, because the Meta values all came from JSON
// originally anyway.
return nil , fmt . Errorf ( "cannot serialize private instance object data: %s" , err )
}
}
var status string
if isOld . Tainted {
status = "tainted"
}
var instKeyRaw interface { }
switch tk := instKey . ( type ) {
case addrs . IntKey :
instKeyRaw = int ( tk )
case addrs . StringKey :
instKeyRaw = string ( tk )
default :
if instKeyRaw != nil {
2019-07-19 02:35:47 +02:00
return nil , fmt . Errorf ( "unsupported instance key: %#v" , instKey )
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
}
}
terraform: Ugly huge change to weave in new State and Plan types
Due to how often the state and plan types are referenced throughout
Terraform, there isn't a great way to switch them out gradually. As a
consequence, this huge commit gets us from the old world to a _compilable_
new world, but still has a large number of known test failures due to
key functionality being stubbed out.
The stubs here are for anything that interacts with providers, since we
now need to do the follow-up work to similarly replace the old
terraform.ResourceProvider interface with its replacement in the new
"providers" package. That work, along with work to fix the remaining
failing tests, will follow in subsequent commits.
The aim here was to replace all references to terraform.State and its
downstream types with states.State, terraform.Plan with plans.Plan,
state.State with statemgr.State, and switch to the new implementations of
the state and plan file formats. However, due to the number of times those
types are used, this also ended up affecting numerous other parts of core
such as terraform.Hook, the backend.Backend interface, and most of the CLI
commands.
Just as with 5861dbf3fc49b19587a31816eb06f511ab861bb4 before, I apologize
in advance to the person who inevitably just found this huge commit while
spelunking through the commit history.
2018-08-14 23:24:45 +02:00
var attributes map [ string ] string
if isOld . Attributes != nil {
attributes = make ( map [ string ] string , len ( isOld . Attributes ) )
for k , v := range isOld . Attributes {
attributes [ k ] = v
}
}
if isOld . ID != "" {
// As a special case, if we don't already have an "id" attribute and
// yet there's a non-empty first-class ID on the old object then we'll
// create a synthetic id attribute to avoid losing that first-class id.
// In practice this generally arises only in tests where state literals
// are hand-written in a non-standard way; real code prior to 0.12
// would always force the first-class ID to be copied into the
// id attribute before storing.
if attributes == nil {
attributes = make ( map [ string ] string , len ( isOld . Attributes ) )
}
if idVal := attributes [ "id" ] ; idVal == "" {
attributes [ "id" ] = isOld . ID
}
}
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
return & instanceObjectStateV4 {
IndexKey : instKeyRaw ,
Status : status ,
Deposed : string ( deposedKey ) ,
terraform: Ugly huge change to weave in new State and Plan types
Due to how often the state and plan types are referenced throughout
Terraform, there isn't a great way to switch them out gradually. As a
consequence, this huge commit gets us from the old world to a _compilable_
new world, but still has a large number of known test failures due to
key functionality being stubbed out.
The stubs here are for anything that interacts with providers, since we
now need to do the follow-up work to similarly replace the old
terraform.ResourceProvider interface with its replacement in the new
"providers" package. That work, along with work to fix the remaining
failing tests, will follow in subsequent commits.
The aim here was to replace all references to terraform.State and its
downstream types with states.State, terraform.Plan with plans.Plan,
state.State with statemgr.State, and switch to the new implementations of
the state and plan file formats. However, due to the number of times those
types are used, this also ended up affecting numerous other parts of core
such as terraform.Hook, the backend.Backend interface, and most of the CLI
commands.
Just as with 5861dbf3fc49b19587a31816eb06f511ab861bb4 before, I apologize
in advance to the person who inevitably just found this huge commit while
spelunking through the commit history.
2018-08-14 23:24:45 +02:00
AttributesFlat : attributes ,
statefile: New package for loading and saving state files
Whereas the parent directory "states" contains the models that represent
state in memory, this package's responsibility is in serializing a subset
of that data to a JSON-based file format and then reloading that data
back into memory later.
For reading, this package supports state file formats going back to
version 1, using lightly-adapted versions of the migration code previously
used in the "terraform" package. State data is upgraded to the latest
version step by step and then transformed into the in-memory state
representation, which is distinct from any of the file format structs in
this package to enable these to evolve separately.
For writing, only the latest version (4) is supported, which is a new
format that is a slightly-flattened version of the new in-memory state
models introduced in the prior commit. This format retains the outputs
from only the root module and it flattens out the module and instance
parts of the hierarchy by including the identifiers for these inside
the child object. The loader then reconstructs the multi-layer structure
we use for more convenient access in memory.
For now, the only testing in this package is of round-tripping different
versions of state through a read and a write, ensuring the output is
as desired. This exercises all of the reading, upgrading, and writing
functions but should be augmented in later commits to improve coverage
and introduce more focused tests for specific parts of the functionality.
2018-06-08 02:35:55 +02:00
SchemaVersion : schemaVersion ,
PrivateRaw : privateJSON ,
} , nil
}
// parseLegacyResourceAddress parses the different identifier format used
// state formats before version 4, like "instance.name.0".
func parseLegacyResourceAddress ( s string ) ( addrs . ResourceInstance , error ) {
var ret addrs . ResourceInstance
// Split based on ".". Every resource address should have at least two
// elements (type and name).
parts := strings . Split ( s , "." )
if len ( parts ) < 2 || len ( parts ) > 4 {
return ret , fmt . Errorf ( "invalid internal resource address format: %s" , s )
}
// Data resource if we have at least 3 parts and the first one is data
ret . Resource . Mode = addrs . ManagedResourceMode
if len ( parts ) > 2 && parts [ 0 ] == "data" {
ret . Resource . Mode = addrs . DataResourceMode
parts = parts [ 1 : ]
}
// If we're not a data resource and we have more than 3, then it is an error
if len ( parts ) > 3 && ret . Resource . Mode != addrs . DataResourceMode {
return ret , fmt . Errorf ( "invalid internal resource address format: %s" , s )
}
// Build the parts of the resource address that are guaranteed to exist
ret . Resource . Type = parts [ 0 ]
ret . Resource . Name = parts [ 1 ]
ret . Key = addrs . NoKey
// If we have more parts, then we have an index. Parse that.
if len ( parts ) > 2 {
idx , err := strconv . ParseInt ( parts [ 2 ] , 0 , 0 )
if err != nil {
return ret , fmt . Errorf ( "error parsing resource address %q: %s" , s , err )
}
ret . Key = addrs . IntKey ( idx )
}
return ret , nil
}
// simplifyImpliedValueType attempts to heuristically simplify a value type
// derived from a legacy stored output value into something simpler that
// is closer to what would've fitted into the pre-v0.12 value type system.
func simplifyImpliedValueType ( ty cty . Type ) cty . Type {
switch {
case ty . IsTupleType ( ) :
// If all of the element types are the same then we'll make this
// a list instead. This is very likely to be true, since prior versions
// of Terraform did not officially support mixed-type collections.
if ty . Equals ( cty . EmptyTuple ) {
// Don't know what the element type would be, then.
return ty
}
etys := ty . TupleElementTypes ( )
ety := etys [ 0 ]
for _ , other := range etys [ 1 : ] {
if ! other . Equals ( ety ) {
// inconsistent types
return ty
}
}
ety = simplifyImpliedValueType ( ety )
return cty . List ( ety )
case ty . IsObjectType ( ) :
// If all of the attribute types are the same then we'll make this
// a map instead. This is very likely to be true, since prior versions
// of Terraform did not officially support mixed-type collections.
if ty . Equals ( cty . EmptyObject ) {
// Don't know what the element type would be, then.
return ty
}
atys := ty . AttributeTypes ( )
var ety cty . Type
for _ , other := range atys {
if ety == cty . NilType {
ety = other
continue
}
if ! other . Equals ( ety ) {
// inconsistent types
return ty
}
}
ety = simplifyImpliedValueType ( ety )
return cty . Map ( ety )
default :
// No other normalizations are possible
return ty
}
}