2015-02-21 20:52:55 +01:00
|
|
|
package remote
|
|
|
|
|
|
|
|
import (
|
|
|
|
"bytes"
|
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
|
|
|
"fmt"
|
2017-05-25 17:01:25 +02:00
|
|
|
"sync"
|
2015-02-21 20:52:55 +01:00
|
|
|
|
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
|
|
|
uuid "github.com/hashicorp/go-uuid"
|
2017-02-15 23:20:59 +01:00
|
|
|
"github.com/hashicorp/terraform/state"
|
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
|
|
|
"github.com/hashicorp/terraform/states"
|
|
|
|
"github.com/hashicorp/terraform/states/statefile"
|
|
|
|
"github.com/hashicorp/terraform/states/statemgr"
|
2015-02-21 20:52:55 +01:00
|
|
|
)
|
|
|
|
|
|
|
|
// State implements the State interfaces in the state package to handle
|
|
|
|
// reading and writing the remote state. This State on its own does no
|
|
|
|
// local caching so every persist will go to the remote storage and local
|
|
|
|
// writes will go to memory.
|
|
|
|
type State struct {
|
2017-05-25 17:01:25 +02:00
|
|
|
mu sync.Mutex
|
|
|
|
|
2015-02-21 20:52:55 +01:00
|
|
|
Client Client
|
|
|
|
|
2020-04-15 22:54:03 +02:00
|
|
|
// We track two pieces of meta data in addition to the state itself:
|
|
|
|
//
|
|
|
|
// lineage - the state's unique ID
|
|
|
|
// serial - the monotonic counter of "versions" of the state
|
|
|
|
//
|
|
|
|
// Both of these (along with state) have a sister field
|
|
|
|
// that represents the values read in from an existing source.
|
|
|
|
// All three of these values are used to determine if the new
|
|
|
|
// state has changed from an existing state we read in.
|
|
|
|
lineage, readLineage string
|
|
|
|
serial, readSerial uint64
|
|
|
|
state, readState *states.State
|
|
|
|
disableLocks bool
|
2015-02-21 20:52:55 +01:00
|
|
|
}
|
|
|
|
|
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 _ statemgr.Full = (*State)(nil)
|
2018-11-13 03:26:49 +01:00
|
|
|
var _ statemgr.Migrator = (*State)(nil)
|
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
|
|
|
|
|
|
|
// statemgr.Reader impl.
|
|
|
|
func (s *State) State() *states.State {
|
2017-05-25 17:01:25 +02:00
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
2015-02-24 06:36:35 +01:00
|
|
|
return s.state.DeepCopy()
|
2015-02-21 20:52:55 +01:00
|
|
|
}
|
|
|
|
|
2018-11-13 03:26:49 +01:00
|
|
|
// StateForMigration is part of our implementation of statemgr.Migrator.
|
|
|
|
func (s *State) StateForMigration() *statefile.File {
|
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
|
|
|
return statefile.New(s.state.DeepCopy(), s.lineage, s.serial)
|
|
|
|
}
|
|
|
|
|
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
|
|
|
// statemgr.Writer impl.
|
|
|
|
func (s *State) WriteState(state *states.State) error {
|
2017-05-25 17:01:25 +02:00
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
2017-07-05 21:34:30 +02:00
|
|
|
// We create a deep copy of the state here, because the caller also has
|
|
|
|
// a reference to the given object and can potentially go on to mutate
|
|
|
|
// it after we return, but we want the snapshot at this point in time.
|
|
|
|
s.state = state.DeepCopy()
|
|
|
|
|
2015-02-21 20:52:55 +01:00
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
2018-11-13 03:26:49 +01:00
|
|
|
// WriteStateForMigration is part of our implementation of statemgr.Migrator.
|
|
|
|
func (s *State) WriteStateForMigration(f *statefile.File, force bool) error {
|
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
2020-04-15 22:54:03 +02:00
|
|
|
// `force` is passed down from the CLI flag and terminates here. Actual
|
|
|
|
// force pushing with the remote backend happens when Put()'ing the contents
|
|
|
|
// in the backend. If force is specified we skip verifications and hand the
|
|
|
|
// context off to the client to use when persitence operations actually take place.
|
|
|
|
c, isForcePusher := s.Client.(ClientForcePusher)
|
|
|
|
if force && isForcePusher {
|
|
|
|
c.EnableForcePush()
|
|
|
|
} else {
|
|
|
|
checkFile := statefile.New(s.state, s.lineage, s.serial)
|
2018-11-13 03:26:49 +01:00
|
|
|
if err := statemgr.CheckValidImport(f, checkFile); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// We create a deep copy of the state here, because the caller also has
|
|
|
|
// a reference to the given object and can potentially go on to mutate
|
|
|
|
// it after we return, but we want the snapshot at this point in time.
|
|
|
|
s.state = f.State.DeepCopy()
|
|
|
|
s.lineage = f.Lineage
|
|
|
|
s.serial = f.Serial
|
|
|
|
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
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
|
|
|
// statemgr.Refresher impl.
|
2015-02-21 20:52:55 +01:00
|
|
|
func (s *State) RefreshState() error {
|
2017-05-25 17:01:25 +02:00
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
2018-10-02 23:55:43 +02:00
|
|
|
return s.refreshState()
|
|
|
|
}
|
2017-05-25 17:01:25 +02:00
|
|
|
|
2018-10-02 23:55:43 +02:00
|
|
|
// refreshState is the main implementation of RefreshState, but split out so
|
|
|
|
// that we can make internal calls to it from methods that are already holding
|
|
|
|
// the s.mu lock.
|
|
|
|
func (s *State) refreshState() error {
|
2015-02-21 20:52:55 +01:00
|
|
|
payload, err := s.Client.Get()
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
2016-07-06 16:48:52 +02:00
|
|
|
// no remote state is OK
|
2016-07-02 00:37:38 +02:00
|
|
|
if payload == nil {
|
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
|
|
|
s.readState = nil
|
|
|
|
s.lineage = ""
|
|
|
|
s.serial = 0
|
2016-07-06 16:48:52 +02:00
|
|
|
return nil
|
2016-07-02 00:37:38 +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
|
|
|
stateFile, err := statefile.Read(bytes.NewReader(payload.Data))
|
2016-07-02 00:37:38 +02:00
|
|
|
if err != nil {
|
|
|
|
return err
|
2015-02-21 20:52:55 +01:00
|
|
|
}
|
|
|
|
|
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
|
|
|
s.lineage = stateFile.Lineage
|
|
|
|
s.serial = stateFile.Serial
|
|
|
|
s.state = stateFile.State
|
2020-04-15 22:54:03 +02:00
|
|
|
|
|
|
|
// Properties from the remote must be separate so we can
|
|
|
|
// track changes as lineage, serial and/or state are mutated
|
|
|
|
s.readLineage = stateFile.Lineage
|
|
|
|
s.readSerial = stateFile.Serial
|
|
|
|
s.readState = s.state.DeepCopy()
|
2015-02-21 20:52:55 +01:00
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
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
|
|
|
// statemgr.Persister impl.
|
2015-02-21 20:52:55 +01:00
|
|
|
func (s *State) PersistState() error {
|
2017-05-25 17:01:25 +02:00
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
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
|
|
|
if s.readState != nil {
|
2020-04-15 22:54:03 +02:00
|
|
|
lineageUnchanged := s.readLineage != "" && s.lineage == s.readLineage
|
|
|
|
serialUnchanged := s.readSerial != 0 && s.serial == s.readSerial
|
|
|
|
stateUnchanged := statefile.StatesMarshalEqual(s.state, s.readState)
|
|
|
|
if stateUnchanged && lineageUnchanged && serialUnchanged {
|
|
|
|
// If the state, lineage or serial haven't changed at all then we have nothing to do.
|
state/remote: Don't persist snapshot for unchanged state
Previously we would write to the backend for every call to PersistState,
even if nothing changed since the last write, but update the serial only
if the state had changed.
The Terraform Cloud & Enterprise state storage have a simple safety check
that any future write with an already-used lineage and serial must be
byte-for-byte identical. StatesMarshalEqual is intended to detect that,
but it only actually detects changes the state itself, and not changes
to the snapshot metadata.
Because we write the current Terraform version into the snapshot metadata
during serialization, we'd previously have an issue where if the first
state write after upgrading Terraform to a new version happened to change
nothing about the state content then we'd write a new snapshot that
differed only by Terraform version, and Terraform Cloud/Enterprise would
then reject it.
The snapshot header is discarded immediately after decoding, so we can't
use information from it when deciding whether to increment the serial.
The next best thing is to skip sending no-op snapshot updates to the state
client in the first place.
These writes are unnecessary anyway, and state storage owners have asked
us in the past to elide these to avoid generating noise in their version
logs, so we'll also finally meet those requests as a nice side-effect of
this change.
We didn't previously have tests for the full flow of retrieving and then
successively updating persisted state snapshots, so this includes a test
which covers that logic and includes an assertion that a no-op update does
not get written to the state client.
2019-06-20 02:42:09 +02:00
|
|
|
return nil
|
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
|
|
|
}
|
state/remote: Don't persist snapshot for unchanged state
Previously we would write to the backend for every call to PersistState,
even if nothing changed since the last write, but update the serial only
if the state had changed.
The Terraform Cloud & Enterprise state storage have a simple safety check
that any future write with an already-used lineage and serial must be
byte-for-byte identical. StatesMarshalEqual is intended to detect that,
but it only actually detects changes the state itself, and not changes
to the snapshot metadata.
Because we write the current Terraform version into the snapshot metadata
during serialization, we'd previously have an issue where if the first
state write after upgrading Terraform to a new version happened to change
nothing about the state content then we'd write a new snapshot that
differed only by Terraform version, and Terraform Cloud/Enterprise would
then reject it.
The snapshot header is discarded immediately after decoding, so we can't
use information from it when deciding whether to increment the serial.
The next best thing is to skip sending no-op snapshot updates to the state
client in the first place.
These writes are unnecessary anyway, and state storage owners have asked
us in the past to elide these to avoid generating noise in their version
logs, so we'll also finally meet those requests as a nice side-effect of
this change.
We didn't previously have tests for the full flow of retrieving and then
successively updating persisted state snapshots, so this includes a test
which covers that logic and includes an assertion that a no-op update does
not get written to the state client.
2019-06-20 02:42:09 +02:00
|
|
|
s.serial++
|
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
|
|
|
} else {
|
|
|
|
// We might be writing a new state altogether, but before we do that
|
|
|
|
// we'll check to make sure there isn't already a snapshot present
|
|
|
|
// that we ought to be updating.
|
2018-10-02 23:55:43 +02:00
|
|
|
err := s.refreshState()
|
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
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("failed checking for existing remote state: %s", err)
|
|
|
|
}
|
|
|
|
if s.lineage == "" { // indicates that no state snapshot is present yet
|
|
|
|
lineage, err := uuid.GenerateUUID()
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("failed to generate initial lineage: %v", err)
|
|
|
|
}
|
|
|
|
s.lineage = lineage
|
|
|
|
s.serial = 0
|
|
|
|
}
|
2017-07-05 21:34:30 +02:00
|
|
|
}
|
2015-02-24 06:30:59 +01:00
|
|
|
|
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
|
|
|
f := statefile.New(s.state, s.lineage, s.serial)
|
|
|
|
|
2015-02-21 20:52:55 +01:00
|
|
|
var buf bytes.Buffer
|
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
|
|
|
err := statefile.Write(f, &buf)
|
|
|
|
if err != nil {
|
2015-02-21 20:52:55 +01:00
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
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
|
|
|
err = s.Client.Put(buf.Bytes())
|
2017-07-05 21:34:30 +02:00
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
|
|
|
// After we've successfully persisted, what we just wrote is our new
|
|
|
|
// reference state until someone calls RefreshState again.
|
2020-04-15 22:54:03 +02:00
|
|
|
// We've potentially overwritten (via force) the state, lineage
|
|
|
|
// and / or serial (and serial was incremented) so we copy over all
|
|
|
|
// three fields so everything matches the new state and a subsequent
|
|
|
|
// operation would correctly detect no changes to the lineage, serial or state.
|
2017-07-05 21:34:30 +02:00
|
|
|
s.readState = s.state.DeepCopy()
|
2020-04-15 22:54:03 +02:00
|
|
|
s.readLineage = s.lineage
|
|
|
|
s.readSerial = s.serial
|
2017-07-05 21:34:30 +02:00
|
|
|
return nil
|
2015-02-21 20:52:55 +01:00
|
|
|
}
|
2017-01-10 00:18:31 +01:00
|
|
|
|
|
|
|
// Lock calls the Client's Lock method if it's implemented.
|
2017-02-15 23:20:59 +01:00
|
|
|
func (s *State) Lock(info *state.LockInfo) (string, error) {
|
2017-05-25 17:01:25 +02:00
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
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
|
|
|
if s.disableLocks {
|
|
|
|
return "", nil
|
|
|
|
}
|
|
|
|
|
2017-02-15 23:20:59 +01:00
|
|
|
if c, ok := s.Client.(ClientLocker); ok {
|
|
|
|
return c.Lock(info)
|
2017-01-10 00:18:31 +01:00
|
|
|
}
|
2017-02-15 23:20:59 +01:00
|
|
|
return "", nil
|
2017-01-10 00:18:31 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
// Unlock calls the Client's Unlock method if it's implemented.
|
2017-02-15 23:20:59 +01:00
|
|
|
func (s *State) Unlock(id string) error {
|
2017-05-25 17:01:25 +02:00
|
|
|
s.mu.Lock()
|
|
|
|
defer s.mu.Unlock()
|
|
|
|
|
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
|
|
|
if s.disableLocks {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
2017-02-15 23:20:59 +01:00
|
|
|
if c, ok := s.Client.(ClientLocker); ok {
|
|
|
|
return c.Unlock(id)
|
2017-01-10 00:18:31 +01:00
|
|
|
}
|
|
|
|
return nil
|
|
|
|
}
|
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
|
|
|
|
|
|
|
// DisableLocks turns the Lock and Unlock methods into no-ops. This is intended
|
|
|
|
// to be called during initialization of a state manager and should not be
|
|
|
|
// called after any of the statemgr.Full interface methods have been called.
|
|
|
|
func (s *State) DisableLocks() {
|
|
|
|
s.disableLocks = true
|
|
|
|
}
|
2018-10-01 23:35:23 +02:00
|
|
|
|
|
|
|
// StateSnapshotMeta returns the metadata from the most recently persisted
|
|
|
|
// or refreshed persistent state snapshot.
|
|
|
|
//
|
|
|
|
// This is an implementation of statemgr.PersistentMeta.
|
|
|
|
func (s *State) StateSnapshotMeta() statemgr.SnapshotMeta {
|
|
|
|
return statemgr.SnapshotMeta{
|
|
|
|
Lineage: s.lineage,
|
|
|
|
Serial: s.serial,
|
|
|
|
}
|
|
|
|
}
|