2018-09-24 18:30:39 +02:00
|
|
|
/*
|
|
|
|
*
|
|
|
|
* Copyright 2017 gRPC authors.
|
|
|
|
*
|
|
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
* you may not use this file except in compliance with the License.
|
|
|
|
* You may obtain a copy of the License at
|
|
|
|
*
|
|
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
*
|
|
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
* See the License for the specific language governing permissions and
|
|
|
|
* limitations under the License.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
package grpc
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
|
|
|
"strings"
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
"sync"
|
|
|
|
"time"
|
2018-09-24 18:30:39 +02:00
|
|
|
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
"google.golang.org/grpc/balancer"
|
|
|
|
"google.golang.org/grpc/credentials"
|
2018-09-24 18:30:39 +02:00
|
|
|
"google.golang.org/grpc/grpclog"
|
2019-02-20 19:51:33 +01:00
|
|
|
"google.golang.org/grpc/internal/channelz"
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
"google.golang.org/grpc/internal/grpcsync"
|
2018-09-24 18:30:39 +02:00
|
|
|
"google.golang.org/grpc/resolver"
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
"google.golang.org/grpc/serviceconfig"
|
2018-09-24 18:30:39 +02:00
|
|
|
)
|
|
|
|
|
|
|
|
// ccResolverWrapper is a wrapper on top of cc for resolvers.
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
// It implements resolver.ClientConn interface.
|
2018-09-24 18:30:39 +02:00
|
|
|
type ccResolverWrapper struct {
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
cc *ClientConn
|
|
|
|
resolverMu sync.Mutex
|
|
|
|
resolver resolver.Resolver
|
|
|
|
done *grpcsync.Event
|
|
|
|
curState resolver.State
|
|
|
|
|
|
|
|
pollingMu sync.Mutex
|
|
|
|
polling chan struct{}
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
// split2 returns the values from strings.SplitN(s, sep, 2).
|
2019-02-20 19:51:33 +01:00
|
|
|
// If sep is not found, it returns ("", "", false) instead.
|
2018-09-24 18:30:39 +02:00
|
|
|
func split2(s, sep string) (string, string, bool) {
|
|
|
|
spl := strings.SplitN(s, sep, 2)
|
|
|
|
if len(spl) < 2 {
|
|
|
|
return "", "", false
|
|
|
|
}
|
|
|
|
return spl[0], spl[1], true
|
|
|
|
}
|
|
|
|
|
|
|
|
// parseTarget splits target into a struct containing scheme, authority and
|
|
|
|
// endpoint.
|
|
|
|
//
|
|
|
|
// If target is not a valid scheme://authority/endpoint, it returns {Endpoint:
|
|
|
|
// target}.
|
|
|
|
func parseTarget(target string) (ret resolver.Target) {
|
|
|
|
var ok bool
|
|
|
|
ret.Scheme, ret.Endpoint, ok = split2(target, "://")
|
|
|
|
if !ok {
|
|
|
|
return resolver.Target{Endpoint: target}
|
|
|
|
}
|
|
|
|
ret.Authority, ret.Endpoint, ok = split2(ret.Endpoint, "/")
|
|
|
|
if !ok {
|
|
|
|
return resolver.Target{Endpoint: target}
|
|
|
|
}
|
|
|
|
return ret
|
|
|
|
}
|
|
|
|
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
// newCCResolverWrapper uses the resolver.Builder to build a Resolver and
|
|
|
|
// returns a ccResolverWrapper object which wraps the newly built resolver.
|
|
|
|
func newCCResolverWrapper(cc *ClientConn, rb resolver.Builder) (*ccResolverWrapper, error) {
|
|
|
|
ccr := &ccResolverWrapper{
|
|
|
|
cc: cc,
|
|
|
|
done: grpcsync.NewEvent(),
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
|
|
|
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
var credsClone credentials.TransportCredentials
|
|
|
|
if creds := cc.dopts.copts.TransportCredentials; creds != nil {
|
|
|
|
credsClone = creds.Clone()
|
|
|
|
}
|
|
|
|
rbo := resolver.BuildOptions{
|
|
|
|
DisableServiceConfig: cc.dopts.disableServiceConfig,
|
|
|
|
DialCreds: credsClone,
|
|
|
|
CredsBundle: cc.dopts.copts.CredsBundle,
|
|
|
|
Dialer: cc.dopts.copts.Dialer,
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
var err error
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
// We need to hold the lock here while we assign to the ccr.resolver field
|
|
|
|
// to guard against a data race caused by the following code path,
|
|
|
|
// rb.Build-->ccr.ReportError-->ccr.poll-->ccr.resolveNow, would end up
|
|
|
|
// accessing ccr.resolver which is being assigned here.
|
|
|
|
ccr.resolverMu.Lock()
|
|
|
|
defer ccr.resolverMu.Unlock()
|
|
|
|
ccr.resolver, err = rb.Build(cc.parsedTarget, ccr, rbo)
|
2018-09-24 18:30:39 +02:00
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
|
|
|
}
|
|
|
|
return ccr, nil
|
|
|
|
}
|
|
|
|
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
func (ccr *ccResolverWrapper) resolveNow(o resolver.ResolveNowOptions) {
|
|
|
|
ccr.resolverMu.Lock()
|
|
|
|
if !ccr.done.HasFired() {
|
|
|
|
ccr.resolver.ResolveNow(o)
|
|
|
|
}
|
|
|
|
ccr.resolverMu.Unlock()
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
func (ccr *ccResolverWrapper) close() {
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
ccr.resolverMu.Lock()
|
2018-09-24 18:30:39 +02:00
|
|
|
ccr.resolver.Close()
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
ccr.done.Fire()
|
|
|
|
ccr.resolverMu.Unlock()
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
|
|
|
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
// poll begins or ends asynchronous polling of the resolver based on whether
|
|
|
|
// err is ErrBadResolverState.
|
|
|
|
func (ccr *ccResolverWrapper) poll(err error) {
|
|
|
|
ccr.pollingMu.Lock()
|
|
|
|
defer ccr.pollingMu.Unlock()
|
|
|
|
if err != balancer.ErrBadResolverState {
|
|
|
|
// stop polling
|
|
|
|
if ccr.polling != nil {
|
|
|
|
close(ccr.polling)
|
|
|
|
ccr.polling = nil
|
|
|
|
}
|
|
|
|
return
|
|
|
|
}
|
|
|
|
if ccr.polling != nil {
|
|
|
|
// already polling
|
|
|
|
return
|
|
|
|
}
|
|
|
|
p := make(chan struct{})
|
|
|
|
ccr.polling = p
|
|
|
|
go func() {
|
|
|
|
for i := 0; ; i++ {
|
|
|
|
ccr.resolveNow(resolver.ResolveNowOptions{})
|
|
|
|
t := time.NewTimer(ccr.cc.dopts.resolveNowBackoff(i))
|
|
|
|
select {
|
|
|
|
case <-p:
|
|
|
|
t.Stop()
|
|
|
|
return
|
|
|
|
case <-ccr.done.Done():
|
|
|
|
// Resolver has been closed.
|
|
|
|
t.Stop()
|
|
|
|
return
|
|
|
|
case <-t.C:
|
|
|
|
select {
|
|
|
|
case <-p:
|
|
|
|
return
|
|
|
|
default:
|
|
|
|
}
|
|
|
|
// Timer expired; re-resolve.
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}()
|
2019-09-09 14:04:58 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
func (ccr *ccResolverWrapper) UpdateState(s resolver.State) {
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
if ccr.done.HasFired() {
|
2019-09-09 14:04:58 +02:00
|
|
|
return
|
|
|
|
}
|
|
|
|
grpclog.Infof("ccResolverWrapper: sending update to cc: %v", s)
|
|
|
|
if channelz.IsOn() {
|
|
|
|
ccr.addChannelzTraceEvent(s)
|
|
|
|
}
|
|
|
|
ccr.curState = s
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
ccr.poll(ccr.cc.updateResolverState(ccr.curState, nil))
|
|
|
|
}
|
|
|
|
|
|
|
|
func (ccr *ccResolverWrapper) ReportError(err error) {
|
|
|
|
if ccr.done.HasFired() {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
grpclog.Warningf("ccResolverWrapper: reporting error to cc: %v", err)
|
|
|
|
if channelz.IsOn() {
|
|
|
|
channelz.AddTraceEvent(ccr.cc.channelzID, &channelz.TraceEventDesc{
|
|
|
|
Desc: fmt.Sprintf("Resolver reported error: %v", err),
|
|
|
|
Severity: channelz.CtWarning,
|
|
|
|
})
|
|
|
|
}
|
|
|
|
ccr.poll(ccr.cc.updateResolverState(resolver.State{}, err))
|
2019-09-09 14:04:58 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
// NewAddress is called by the resolver implementation to send addresses to gRPC.
|
2018-09-24 18:30:39 +02:00
|
|
|
func (ccr *ccResolverWrapper) NewAddress(addrs []resolver.Address) {
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
if ccr.done.HasFired() {
|
2019-02-20 19:51:33 +01:00
|
|
|
return
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
2019-02-20 19:51:33 +01:00
|
|
|
grpclog.Infof("ccResolverWrapper: sending new addresses to cc: %v", addrs)
|
|
|
|
if channelz.IsOn() {
|
2019-09-09 14:04:58 +02:00
|
|
|
ccr.addChannelzTraceEvent(resolver.State{Addresses: addrs, ServiceConfig: ccr.curState.ServiceConfig})
|
2019-02-20 19:51:33 +01:00
|
|
|
}
|
2019-09-09 14:04:58 +02:00
|
|
|
ccr.curState.Addresses = addrs
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
ccr.poll(ccr.cc.updateResolverState(ccr.curState, nil))
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
|
|
|
|
2019-09-09 14:04:58 +02:00
|
|
|
// NewServiceConfig is called by the resolver implementation to send service
|
2018-09-24 18:30:39 +02:00
|
|
|
// configs to gRPC.
|
|
|
|
func (ccr *ccResolverWrapper) NewServiceConfig(sc string) {
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
if ccr.done.HasFired() {
|
2019-02-20 19:51:33 +01:00
|
|
|
return
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|
2019-02-20 19:51:33 +01:00
|
|
|
grpclog.Infof("ccResolverWrapper: got new service config: %v", sc)
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
if ccr.cc.dopts.disableServiceConfig {
|
|
|
|
grpclog.Infof("Service config lookups disabled; ignoring config")
|
|
|
|
return
|
|
|
|
}
|
|
|
|
scpr := parseServiceConfig(sc)
|
|
|
|
if scpr.Err != nil {
|
|
|
|
grpclog.Warningf("ccResolverWrapper: error parsing service config: %v", scpr.Err)
|
|
|
|
if channelz.IsOn() {
|
|
|
|
channelz.AddTraceEvent(ccr.cc.channelzID, &channelz.TraceEventDesc{
|
|
|
|
Desc: fmt.Sprintf("Error parsing service config: %v", scpr.Err),
|
|
|
|
Severity: channelz.CtWarning,
|
|
|
|
})
|
|
|
|
}
|
|
|
|
ccr.poll(balancer.ErrBadResolverState)
|
|
|
|
return
|
|
|
|
}
|
2019-09-09 14:04:58 +02:00
|
|
|
if channelz.IsOn() {
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
ccr.addChannelzTraceEvent(resolver.State{Addresses: ccr.curState.Addresses, ServiceConfig: scpr})
|
2019-09-09 14:04:58 +02:00
|
|
|
}
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
ccr.curState.ServiceConfig = scpr
|
|
|
|
ccr.poll(ccr.cc.updateResolverState(ccr.curState, nil))
|
|
|
|
}
|
|
|
|
|
|
|
|
func (ccr *ccResolverWrapper) ParseServiceConfig(scJSON string) *serviceconfig.ParseResult {
|
|
|
|
return parseServiceConfig(scJSON)
|
2019-02-20 19:51:33 +01:00
|
|
|
}
|
|
|
|
|
2019-09-09 14:04:58 +02:00
|
|
|
func (ccr *ccResolverWrapper) addChannelzTraceEvent(s resolver.State) {
|
|
|
|
var updates []string
|
command: Unmanaged providers
This adds supports for "unmanaged" providers, or providers with process
lifecycles not controlled by Terraform. These providers are assumed to
be started before Terraform is launched, and are assumed to shut
themselves down after Terraform has finished running.
To do this, we must update the go-plugin dependency to v1.3.0, which
added support for the "test mode" plugin serving that powers all this.
As a side-effect of not needing to manage the process lifecycle anymore,
Terraform also no longer needs to worry about the provider's binary, as
it won't be used for anything anymore. Because of this, we can disable
the init behavior that concerns itself with downloading that provider's
binary, checking its version, and otherwise managing the binary.
This is all managed on a per-provider basis, so managed providers that
Terraform downloads, starts, and stops can be used in the same commands
as unmanaged providers. The TF_REATTACH_PROVIDERS environment variable
is added, and is a JSON encoding of the provider's address to the
information we need to connect to it.
This change enables two benefits: first, delve and other debuggers can
now be attached to provider server processes, and Terraform can connect.
This allows for attaching debuggers to provider processes, which before
was difficult to impossible. Second, it allows the SDK test framework to
host the provider in the same process as the test driver, while running
a production Terraform binary against the provider. This allows for Go's
built-in race detector and test coverage tooling to work as expected in
provider tests.
Unmanaged providers are expected to work in the exact same way as
managed providers, with one caveat: Terraform kills provider processes
and restarts them once per graph walk, meaning multiple times during
most Terraform CLI commands. As unmanaged providers can't be killed by
Terraform, and have no visibility into graph walks, unmanaged providers
are likely to have differences in how their global mutable state behaves
when compared to managed providers. Namely, unmanaged providers are
likely to retain global state when managed providers would have reset
it. Developers relying on global state should be aware of this.
2020-05-27 02:48:57 +02:00
|
|
|
var oldSC, newSC *ServiceConfig
|
|
|
|
var oldOK, newOK bool
|
|
|
|
if ccr.curState.ServiceConfig != nil {
|
|
|
|
oldSC, oldOK = ccr.curState.ServiceConfig.Config.(*ServiceConfig)
|
|
|
|
}
|
|
|
|
if s.ServiceConfig != nil {
|
|
|
|
newSC, newOK = s.ServiceConfig.Config.(*ServiceConfig)
|
|
|
|
}
|
|
|
|
if oldOK != newOK || (oldOK && newOK && oldSC.rawJSONString != newSC.rawJSONString) {
|
2019-09-09 14:04:58 +02:00
|
|
|
updates = append(updates, "service config updated")
|
|
|
|
}
|
|
|
|
if len(ccr.curState.Addresses) > 0 && len(s.Addresses) == 0 {
|
|
|
|
updates = append(updates, "resolver returned an empty address list")
|
|
|
|
} else if len(ccr.curState.Addresses) == 0 && len(s.Addresses) > 0 {
|
|
|
|
updates = append(updates, "resolver returned new addresses")
|
2019-02-20 19:51:33 +01:00
|
|
|
}
|
2019-09-09 14:04:58 +02:00
|
|
|
channelz.AddTraceEvent(ccr.cc.channelzID, &channelz.TraceEventDesc{
|
|
|
|
Desc: fmt.Sprintf("Resolver state updated: %+v (%v)", s, strings.Join(updates, "; ")),
|
|
|
|
Severity: channelz.CtINFO,
|
|
|
|
})
|
2018-09-24 18:30:39 +02:00
|
|
|
}
|