consul/internal/storage/conformance/conformance.go

659 lines
17 KiB
Go
Raw Normal View History

// Copyright (c) HashiCorp, Inc.
[COMPLIANCE] License changes (#18443) * Adding explicit MPL license for sub-package This directory and its subdirectories (packages) contain files licensed with the MPLv2 `LICENSE` file in this directory and are intentionally licensed separately from the BSL `LICENSE` file at the root of this repository. * Adding explicit MPL license for sub-package This directory and its subdirectories (packages) contain files licensed with the MPLv2 `LICENSE` file in this directory and are intentionally licensed separately from the BSL `LICENSE` file at the root of this repository. * Updating the license from MPL to Business Source License Going forward, this project will be licensed under the Business Source License v1.1. Please see our blog post for more details at <Blog URL>, FAQ at www.hashicorp.com/licensing-faq, and details of the license at www.hashicorp.com/bsl. * add missing license headers * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 * Update copyright file headers to BUSL-1.1 --------- Co-authored-by: hashicorp-copywrite[bot] <110428419+hashicorp-copywrite[bot]@users.noreply.github.com>
2023-08-11 13:12:13 +00:00
// SPDX-License-Identifier: BUSL-1.1
package conformance
import (
"context"
"errors"
"fmt"
"math/rand"
"testing"
"time"
"github.com/stretchr/testify/require"
"google.golang.org/protobuf/proto"
"google.golang.org/protobuf/testing/protocmp"
"github.com/hashicorp/consul/internal/storage"
"github.com/hashicorp/consul/proto-public/pbresource"
"github.com/hashicorp/consul/proto/private/prototest"
"github.com/hashicorp/consul/sdk/testutil/retry"
)
type TestOptions struct {
// NewBackend will be called to construct a storage.Backend to run the tests
// against.
NewBackend func(t *testing.T) storage.Backend
// SupportsStronglyConsistentList indicates whether the given storage backend
// supports strongly consistent list operations.
SupportsStronglyConsistentList bool
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
// IgnoreWatchListSnapshotOperations indicates whether a given storage
// backend is expected to be consistent enough with reads to emit
// WatchEvent_Upsert after the initial sync.
//
// For instance, a replicated copy of the state store will have stale data
// and may return an initial snapshot of nothing, and follow it up by an
// upsert.
IgnoreWatchListSnapshotOperations bool
}
// Test runs a suite of tests against a storage.Backend implementation to check
// it correctly implements our required behaviours.
func Test(t *testing.T, opts TestOptions) {
require.NotNil(t, opts.NewBackend, "NewBackend method is required")
t.Run("Read", func(t *testing.T) { testRead(t, opts) })
t.Run("CAS Write", func(t *testing.T) { testCASWrite(t, opts) })
t.Run("CAS Delete", func(t *testing.T) { testCASDelete(t, opts) })
t.Run("ListByOwner", func(t *testing.T) { testListByOwner(t, opts) })
testListWatch(t, opts)
}
func testRead(t *testing.T, opts TestOptions) {
ctx := testContext(t)
for consistency, check := range map[storage.ReadConsistency]consistencyChecker{
storage.EventualConsistency: eventually,
storage.StrongConsistency: immediately,
} {
t.Run(consistency.String(), func(t *testing.T) {
res := &pbresource.Resource{
Id: &pbresource.ID{
Type: typeAv1,
Tenancy: tenancyDefault,
Name: "web",
Uid: "a",
},
}
t.Run("simple", func(t *testing.T) {
backend := opts.NewBackend(t)
_, err := backend.WriteCAS(ctx, res)
require.NoError(t, err)
check(t, func(t testingT) {
output, err := backend.Read(ctx, consistency, res.Id)
require.NoError(t, err)
prototest.AssertDeepEqual(t, res, output, ignoreVersion)
})
})
t.Run("no uid", func(t *testing.T) {
backend := opts.NewBackend(t)
_, err := backend.WriteCAS(ctx, res)
require.NoError(t, err)
id := clone(res.Id)
id.Uid = ""
check(t, func(t testingT) {
output, err := backend.Read(ctx, consistency, id)
require.NoError(t, err)
prototest.AssertDeepEqual(t, res, output, ignoreVersion)
})
})
t.Run("different id", func(t *testing.T) {
backend := opts.NewBackend(t)
_, err := backend.WriteCAS(ctx, res)
require.NoError(t, err)
id := clone(res.Id)
id.Name = "different"
check(t, func(t testingT) {
_, err := backend.Read(ctx, consistency, id)
require.ErrorIs(t, err, storage.ErrNotFound)
})
})
t.Run("different uid", func(t *testing.T) {
backend := opts.NewBackend(t)
_, err := backend.WriteCAS(ctx, res)
require.NoError(t, err)
id := clone(res.Id)
id.Uid = "b"
check(t, func(t testingT) {
_, err := backend.Read(ctx, consistency, id)
require.ErrorIs(t, err, storage.ErrNotFound)
})
})
t.Run("different GroupVersion", func(t *testing.T) {
backend := opts.NewBackend(t)
_, err := backend.WriteCAS(ctx, res)
require.NoError(t, err)
id := clone(res.Id)
id.Type = typeAv2
check(t, func(t testingT) {
_, err := backend.Read(ctx, consistency, id)
require.Error(t, err)
var e storage.GroupVersionMismatchError
if errors.As(err, &e) {
2023-04-04 16:30:06 +00:00
prototest.AssertDeepEqual(t, id.Type, e.RequestedType)
prototest.AssertDeepEqual(t, res, e.Stored, ignoreVersion)
} else {
t.Fatalf("expected storage.GroupVersionMismatchError, got: %T", err)
}
})
})
})
}
}
func testCASWrite(t *testing.T, opts TestOptions) {
t.Run("version-based CAS", func(t *testing.T) {
backend := opts.NewBackend(t)
ctx := testContext(t)
v1 := &pbresource.Resource{
Id: &pbresource.ID{
Type: typeB,
Tenancy: tenancyDefault,
Name: "web",
Uid: "a",
},
}
v1.Version = "some-version"
_, err := backend.WriteCAS(ctx, v1)
require.ErrorIs(t, err, storage.ErrCASFailure)
v1.Version = ""
v1, err = backend.WriteCAS(ctx, v1)
require.NoError(t, err)
require.NotEmpty(t, v1.Version)
v2, err := backend.WriteCAS(ctx, v1)
require.NoError(t, err)
require.NotEmpty(t, v2.Version)
require.NotEqual(t, v1.Version, v2.Version)
v3 := clone(v2)
v3.Version = ""
_, err = backend.WriteCAS(ctx, v3)
require.ErrorIs(t, err, storage.ErrCASFailure)
v3.Version = v1.Version
_, err = backend.WriteCAS(ctx, v3)
require.ErrorIs(t, err, storage.ErrCASFailure)
})
t.Run("uid immutability", func(t *testing.T) {
backend := opts.NewBackend(t)
ctx := testContext(t)
v1, err := backend.WriteCAS(ctx, &pbresource.Resource{
Id: &pbresource.ID{
Type: typeB,
Tenancy: tenancyDefault,
Name: "web",
Uid: "a",
},
})
require.NoError(t, err)
// Uid cannot change.
v2 := clone(v1)
v2.Id.Uid = ""
_, err = backend.WriteCAS(ctx, v2)
require.Error(t, err)
v2.Id.Uid = "b"
_, err = backend.WriteCAS(ctx, v2)
require.ErrorIs(t, err, storage.ErrWrongUid)
v2.Id.Uid = v1.Id.Uid
v2, err = backend.WriteCAS(ctx, v2)
require.NoError(t, err)
// Uid can change after original resource is deleted.
require.NoError(t, backend.DeleteCAS(ctx, v2.Id, v2.Version))
v3 := clone(v2)
v3.Id.Uid = "b"
v3.Version = ""
_, err = backend.WriteCAS(ctx, v3)
require.NoError(t, err)
})
}
func testCASDelete(t *testing.T, opts TestOptions) {
t.Run("version-based CAS", func(t *testing.T) {
backend := opts.NewBackend(t)
ctx := testContext(t)
res, err := backend.WriteCAS(ctx, &pbresource.Resource{
Id: &pbresource.ID{
Type: typeB,
Tenancy: tenancyDefault,
Name: "web",
Uid: "a",
},
})
require.NoError(t, err)
require.ErrorIs(t, backend.DeleteCAS(ctx, res.Id, ""), storage.ErrCASFailure)
require.ErrorIs(t, backend.DeleteCAS(ctx, res.Id, "some-version"), storage.ErrCASFailure)
require.NoError(t, backend.DeleteCAS(ctx, res.Id, res.Version))
eventually(t, func(t testingT) {
_, err = backend.Read(ctx, storage.EventualConsistency, res.Id)
require.ErrorIs(t, err, storage.ErrNotFound)
})
})
t.Run("uid must match", func(t *testing.T) {
backend := opts.NewBackend(t)
ctx := testContext(t)
res, err := backend.WriteCAS(ctx, &pbresource.Resource{
Id: &pbresource.ID{
Type: typeB,
Tenancy: tenancyDefault,
Name: "web",
Uid: "a",
},
})
require.NoError(t, err)
id := clone(res.Id)
id.Uid = "b"
require.NoError(t, backend.DeleteCAS(ctx, id, res.Version))
eventually(t, func(t testingT) {
_, err = backend.Read(ctx, storage.EventualConsistency, res.Id)
require.NoError(t, err)
})
})
}
func testListWatch(t *testing.T, opts TestOptions) {
testCases := map[string]struct {
resourceType storage.UnversionedType
tenancy *pbresource.Tenancy
namePrefix string
results []*pbresource.Resource
}{
"simple #1": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: tenancyDefault,
namePrefix: "",
results: []*pbresource.Resource{
seedData[0],
seedData[1],
seedData[2],
},
},
"simple #2": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: tenancyOther,
namePrefix: "",
results: []*pbresource.Resource{
seedData[3],
},
},
"fixed tenancy, name prefix": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: tenancyDefault,
namePrefix: "a",
results: []*pbresource.Resource{
seedData[0],
seedData[1],
},
},
"wildcard tenancy": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: &pbresource.Tenancy{
Partition: storage.Wildcard,
Namespace: storage.Wildcard,
},
namePrefix: "",
results: []*pbresource.Resource{
seedData[0],
seedData[1],
seedData[2],
seedData[3],
seedData[5],
},
},
"fixed partition, wildcard namespace": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: &pbresource.Tenancy{
Partition: "default",
Namespace: storage.Wildcard,
},
namePrefix: "",
results: []*pbresource.Resource{
seedData[0],
seedData[1],
seedData[2],
seedData[5],
},
},
"wildcard partition, fixed namespace": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: &pbresource.Tenancy{
Partition: storage.Wildcard,
Namespace: "default",
},
namePrefix: "",
results: []*pbresource.Resource{
seedData[0],
seedData[1],
seedData[2],
},
},
"wildcard tenancy, name prefix": {
resourceType: storage.UnversionedTypeFrom(typeAv1),
tenancy: &pbresource.Tenancy{
Partition: storage.Wildcard,
Namespace: storage.Wildcard,
},
namePrefix: "a",
results: []*pbresource.Resource{
seedData[0],
seedData[1],
seedData[3],
seedData[5],
},
},
// TODO(peering/v2) add tests for peer tenancy
}
t.Run("List", func(t *testing.T) {
ctx := testContext(t)
consistencyModes := map[storage.ReadConsistency]consistencyChecker{
storage.EventualConsistency: eventually,
}
if opts.SupportsStronglyConsistentList {
consistencyModes[storage.StrongConsistency] = immediately
}
for consistency, check := range consistencyModes {
t.Run(consistency.String(), func(t *testing.T) {
for desc, tc := range testCases {
t.Run(desc, func(t *testing.T) {
backend := opts.NewBackend(t)
for _, r := range seedData {
_, err := backend.WriteCAS(ctx, r)
require.NoError(t, err)
}
check(t, func(t testingT) {
res, err := backend.List(ctx, consistency, tc.resourceType, tc.tenancy, tc.namePrefix)
require.NoError(t, err)
prototest.AssertElementsMatch(t, res, tc.results, ignoreVersion)
})
})
}
})
}
})
t.Run("WatchList", func(t *testing.T) {
for desc, tc := range testCases {
t.Run(fmt.Sprintf("%s - initial snapshot", desc), func(t *testing.T) {
backend := opts.NewBackend(t)
ctx := testContext(t)
// Write the seed data before the watch has been established.
for _, r := range seedData {
_, err := backend.WriteCAS(ctx, r)
require.NoError(t, err)
}
watch, err := backend.WatchList(ctx, tc.resourceType, tc.tenancy, tc.namePrefix)
require.NoError(t, err)
t.Cleanup(watch.Close)
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
expectNum := len(tc.results) + 1
for i := 0; i < expectNum; i++ {
ctx, cancel := context.WithTimeout(ctx, 5*time.Second)
t.Cleanup(cancel)
event, err := watch.Next(ctx)
require.NoError(t, err)
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
if opts.IgnoreWatchListSnapshotOperations && event.GetEndOfSnapshot() != nil {
continue // ignore
} else if !opts.IgnoreWatchListSnapshotOperations && i == expectNum-1 {
require.NotNil(t, event.GetEndOfSnapshot(), "expected EndOfSnapshot got %T", event.GetEvent())
continue
}
require.NotNil(t, event.GetUpsert(), "index=%d", i)
prototest.AssertContainsElement(t, tc.results, event.GetUpsert().Resource, ignoreVersion)
}
})
t.Run(fmt.Sprintf("%s - following events", desc), func(t *testing.T) {
backend := opts.NewBackend(t)
ctx := testContext(t)
watch, err := backend.WatchList(ctx, tc.resourceType, tc.tenancy, tc.namePrefix)
require.NoError(t, err)
t.Cleanup(watch.Close)
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
{ // read snapshot end
ctx, cancel := context.WithTimeout(ctx, 5*time.Second)
t.Cleanup(cancel)
event, err := watch.Next(ctx)
require.NoError(t, err)
require.NotNil(t, event.GetEndOfSnapshot())
}
// Write the seed data after the watch has been established.
for _, r := range seedData {
_, err := backend.WriteCAS(ctx, r)
require.NoError(t, err)
}
for i := 0; i < len(tc.results); i++ {
ctx, cancel := context.WithTimeout(ctx, 5*time.Second)
t.Cleanup(cancel)
event, err := watch.Next(ctx)
require.NoError(t, err)
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
require.NotNil(t, event.GetUpsert())
prototest.AssertContainsElement(t, tc.results, event.GetUpsert().Resource, ignoreVersion)
// Check that Read implements "monotonic reads" with Watch.
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
readRes, err := backend.Read(ctx, storage.EventualConsistency, event.GetUpsert().Resource.Id)
require.NoError(t, err)
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
prototest.AssertDeepEqual(t, event.GetUpsert().Resource, readRes)
}
// Delete a random resource to check we get an event.
del, err := backend.Read(ctx, storage.EventualConsistency, tc.results[rand.Intn(len(tc.results))].Id)
require.NoError(t, err)
require.NoError(t, backend.DeleteCAS(ctx, del.Id, del.Version))
ctx, cancel := context.WithTimeout(ctx, 5*time.Second)
t.Cleanup(cancel)
event, err := watch.Next(ctx)
require.NoError(t, err)
v2: ensure the controller caches are fully populated before first use (#20421) The new controller caches are initialized before the DependencyMappers or the Reconciler run, but importantly they are not populated. The expectation is that when the WatchList call is made to the resource service it will send an initial snapshot of all resources matching a single type, and then perpetually send UPSERT/DELETE events afterward. This initial snapshot will cycle through the caching layer and will catch it up to reflect the stored data. Critically the dependency mappers and reconcilers will race against the restoration of the caches on server startup or leader election. During this time it is possible a mapper or reconciler will use the cache to lookup a specific relationship and not find it. That very same reconciler may choose to then recompute some persisted resource and in effect rewind it to a prior computed state. Change - Since we are updating the behavior of the WatchList RPC, it was aligned to match that of pbsubscribe and pbpeerstream using a protobuf oneof instead of the enum+fields option. - The WatchList rpc now has 3 alternating response events: Upsert, Delete, EndOfSnapshot. When set the initial batch of "snapshot" Upserts sent on a new watch, those operations will be followed by an EndOfSnapshot event before beginning the never-ending sequence of Upsert/Delete events. - Within the Controller startup code we will launch N+1 goroutines to execute WatchList queries for the watched types. The UPSERTs will be applied to the nascent cache only (no mappers will execute). - Upon witnessing the END operation, those goroutines will terminate. - When all cache priming routines complete, then the normal set of N+1 long lived watch routines will launch to officially witness all events in the system using the primed cached.
2024-02-02 21:11:05 +00:00
require.NotNil(t, event.GetDelete())
prototest.AssertDeepEqual(t, del, event.GetDelete().Resource)
// Check that Read implements "monotonic reads" with Watch.
_, err = backend.Read(ctx, storage.EventualConsistency, del.Id)
require.ErrorIs(t, err, storage.ErrNotFound)
})
}
})
}
func testListByOwner(t *testing.T, opts TestOptions) {
backend := opts.NewBackend(t)
ctx := testContext(t)
owner, err := backend.WriteCAS(ctx, &pbresource.Resource{
Id: &pbresource.ID{
Type: typeAv1,
Tenancy: tenancyDefault,
Name: "owner",
Uid: "a",
},
})
require.NoError(t, err)
r1, err := backend.WriteCAS(ctx, &pbresource.Resource{
Id: &pbresource.ID{
Type: typeB,
Tenancy: tenancyDefault,
Name: "r1",
Uid: "a",
},
Owner: owner.Id,
})
require.NoError(t, err)
r2, err := backend.WriteCAS(ctx, &pbresource.Resource{
Id: &pbresource.ID{
Type: typeAv2,
Tenancy: tenancyDefault,
Name: "r2",
Uid: "a",
},
Owner: owner.Id,
})
require.NoError(t, err)
eventually(t, func(t testingT) {
res, err := backend.ListByOwner(ctx, owner.Id)
require.NoError(t, err)
prototest.AssertElementsMatch(t, res, []*pbresource.Resource{r1, r2})
})
t.Run("references are anchored to a specific uid", func(t *testing.T) {
id := clone(owner.Id)
id.Uid = "different"
eventually(t, func(t testingT) {
res, err := backend.ListByOwner(ctx, id)
require.NoError(t, err)
require.Empty(t, res)
})
})
t.Run("deleting the owner doesn't remove the references", func(t *testing.T) {
require.NoError(t, backend.DeleteCAS(ctx, owner.Id, owner.Version))
eventually(t, func(t testingT) {
res, err := backend.ListByOwner(ctx, owner.Id)
require.NoError(t, err)
prototest.AssertElementsMatch(t, res, []*pbresource.Resource{r1, r2})
})
})
t.Run("deleting the owned resource removes its reference", func(t *testing.T) {
require.NoError(t, backend.DeleteCAS(ctx, r2.Id, r2.Version))
eventually(t, func(t testingT) {
res, err := backend.ListByOwner(ctx, owner.Id)
require.NoError(t, err)
prototest.AssertElementsMatch(t, res, []*pbresource.Resource{r1})
})
})
}
var (
typeAv1 = &pbresource.Type{
Group: "test",
GroupVersion: "v1",
Kind: "a",
}
typeAv2 = &pbresource.Type{
Group: "test",
GroupVersion: "v2",
Kind: "a",
}
typeB = &pbresource.Type{
Group: "test",
GroupVersion: "v1",
Kind: "b",
}
tenancyDefault = &pbresource.Tenancy{
Partition: "default",
Namespace: "default",
}
tenancyDefaultOtherNamespace = &pbresource.Tenancy{
Partition: "default",
Namespace: "other",
}
tenancyOther = &pbresource.Tenancy{
Partition: "billing",
Namespace: "payments",
}
seedData = []*pbresource.Resource{
resource(typeAv1, tenancyDefault, "admin"), // 0
resource(typeAv1, tenancyDefault, "api"), // 1
resource(typeAv2, tenancyDefault, "web"), // 2
resource(typeAv1, tenancyOther, "api"), // 3
resource(typeB, tenancyDefault, "admin"), // 4
resource(typeAv1, tenancyDefaultOtherNamespace, "autoscaler"), // 5
}
ignoreVersion = protocmp.IgnoreFields(&pbresource.Resource{}, "version")
)
func resource(typ *pbresource.Type, ten *pbresource.Tenancy, name string) *pbresource.Resource {
return &pbresource.Resource{
Id: &pbresource.ID{
Type: typ,
Tenancy: ten,
Name: name,
Uid: "a",
},
}
}
func testContext(t *testing.T) context.Context {
ctx, cancel := context.WithCancel(context.Background())
t.Cleanup(cancel)
return ctx
}
func clone[T proto.Message](v T) T { return proto.Clone(v).(T) }
type testingT interface {
require.TestingT
prototest.TestingT
}
type consistencyChecker func(t *testing.T, fn func(testingT))
func eventually(t *testing.T, fn func(testingT)) {
t.Helper()
retry.Run(t, func(r *retry.R) { fn(r) })
}
func immediately(t *testing.T, fn func(testingT)) {
t.Helper()
fn(t)
}