|
| 1 | +// Copyright (c) 2025 Uber Technologies, Inc. |
| 2 | +// |
| 3 | +// Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | +// you may not use this file except in compliance with the License. |
| 5 | +// You may obtain a copy of the License at |
| 6 | +// |
| 7 | +// http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | +// |
| 9 | +// Unless required by applicable law or agreed to in writing, software |
| 10 | +// distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 | +// See the License for the specific language governing permissions and |
| 13 | +// limitations under the License. |
| 14 | + |
| 15 | +// Package record holds the record-stage queue controller. It consumes Record |
| 16 | +// messages (a request id) published by buildsignal once a build reaches a |
| 17 | +// terminal status, and turns that outcome into durable validation state. See |
| 18 | +// doc/rfc/stovepipe/steps/record.md. |
| 19 | +// |
| 20 | +// Phase 1 records only the queue's last-green bookmark, which process reads to |
| 21 | +// choose an incremental build baseline. Validation facts and downstream hooks |
| 22 | +// are not implemented yet. |
| 23 | +package record |
| 24 | + |
| 25 | +import ( |
| 26 | + "context" |
| 27 | + "errors" |
| 28 | + "fmt" |
| 29 | + |
| 30 | + "github.com/uber-go/tally" |
| 31 | + "github.com/uber/submitqueue/platform/consumer" |
| 32 | + "github.com/uber/submitqueue/platform/metrics" |
| 33 | + "github.com/uber/submitqueue/stovepipe/core/loader" |
| 34 | + stovepipemq "github.com/uber/submitqueue/stovepipe/core/messagequeue" |
| 35 | + "github.com/uber/submitqueue/stovepipe/entity" |
| 36 | + "github.com/uber/submitqueue/stovepipe/extension/storage" |
| 37 | + "go.uber.org/zap" |
| 38 | +) |
| 39 | + |
| 40 | +// Controller consumes Record messages and advances the queue's last-green |
| 41 | +// bookmark when the request's build succeeded. Implements consumer.Controller. |
| 42 | +type Controller struct { |
| 43 | + logger *zap.SugaredLogger |
| 44 | + metricsScope tally.Scope |
| 45 | + stores storage.Factory |
| 46 | + topicKey consumer.TopicKey |
| 47 | + consumerGroup string |
| 48 | +} |
| 49 | + |
| 50 | +// Verify Controller implements consumer.Controller interface at compile time. |
| 51 | +var _ consumer.Controller = (*Controller)(nil) |
| 52 | + |
| 53 | +// _opName is the metric operation name shared by every emit in this file. |
| 54 | +const _opName = "record" |
| 55 | + |
| 56 | +// NewController creates a new record controller. |
| 57 | +func NewController( |
| 58 | + logger *zap.SugaredLogger, |
| 59 | + scope tally.Scope, |
| 60 | + stores storage.Factory, |
| 61 | + topicKey consumer.TopicKey, |
| 62 | + consumerGroup string, |
| 63 | +) *Controller { |
| 64 | + return &Controller{ |
| 65 | + logger: logger.Named("record_controller"), |
| 66 | + metricsScope: scope.SubScope("record_controller"), |
| 67 | + stores: stores, |
| 68 | + topicKey: topicKey, |
| 69 | + consumerGroup: consumerGroup, |
| 70 | + } |
| 71 | +} |
| 72 | + |
| 73 | +// Process loads the request referenced by the delivery and, when its build |
| 74 | +// succeeded, advances the queue's last-green bookmark. Returns nil to ack |
| 75 | +// (success) or an error to nack (retry) / reject (DLQ). |
| 76 | +// |
| 77 | +// buildsignal stamps the outcome on the request before publishing here, so a |
| 78 | +// request without a build outcome is a producer invariant violation rather |
| 79 | +// than a state this stage waits for. |
| 80 | +func (c *Controller) Process(ctx context.Context, delivery consumer.Delivery) error { |
| 81 | + msg := delivery.Message() |
| 82 | + |
| 83 | + rec := &stovepipemq.Record{} |
| 84 | + if err := stovepipemq.Unmarshal(msg.Payload, rec); err != nil { |
| 85 | + metrics.NamedCounter(c.metricsScope, _opName, "deserialize_errors", 1) |
| 86 | + // Non-retryable: a malformed message will never succeed regardless of retries. |
| 87 | + return fmt.Errorf("failed to deserialize record: %w", err) |
| 88 | + } |
| 89 | + |
| 90 | + store, err := c.stores.For(storage.Config{QueueName: rec.GetQueueName()}) |
| 91 | + if err != nil { |
| 92 | + metrics.NamedCounter(c.metricsScope, _opName, "storage_resolve_errors", 1) |
| 93 | + // Non-retryable: a missing or unresolvable queue is a malformed message. |
| 94 | + return fmt.Errorf("failed to resolve storage for queue %q: %w", rec.GetQueueName(), err) |
| 95 | + } |
| 96 | + |
| 97 | + request, err := c.loadRequest(ctx, store, rec.Id) |
| 98 | + if err != nil { |
| 99 | + metrics.NamedCounter(c.metricsScope, _opName, "storage_errors", 1) |
| 100 | + return err |
| 101 | + } |
| 102 | + |
| 103 | + // The payload's queue must match the request's authoritative queue; a |
| 104 | + // mismatch is a malformed message. Non-retryable — reject to the DLQ. |
| 105 | + if rec.GetQueueName() != "" && rec.GetQueueName() != request.Queue { |
| 106 | + metrics.NamedCounter(c.metricsScope, _opName, "queue_mismatch", 1) |
| 107 | + return fmt.Errorf("payload queue %q does not match queue %q of request %s", rec.GetQueueName(), request.Queue, request.ID) |
| 108 | + } |
| 109 | + |
| 110 | + switch request.State { |
| 111 | + case entity.RequestStateSucceeded: |
| 112 | + if err := c.advanceLastGreen(ctx, store, request); err != nil { |
| 113 | + metrics.NamedCounter(c.metricsScope, _opName, "storage_errors", 1) |
| 114 | + return err |
| 115 | + } |
| 116 | + return nil |
| 117 | + |
| 118 | + case entity.RequestStateFailed, entity.RequestStateCancelled: |
| 119 | + // A verdict, but not a green one: nothing to record in phase 1. A |
| 120 | + // cancelled build decided nothing about the commit at all. |
| 121 | + metrics.NamedCounter(c.metricsScope, _opName, "not_green", 1) |
| 122 | + return nil |
| 123 | + |
| 124 | + case entity.RequestStateSuperseded: |
| 125 | + // Terminal without a build outcome. buildsignal never publishes for a |
| 126 | + // superseded request, so this is unreachable in practice. |
| 127 | + metrics.NamedCounter(c.metricsScope, _opName, "superseded", 1) |
| 128 | + return nil |
| 129 | + |
| 130 | + default: |
| 131 | + // Non-retryable: buildsignal publishes only after committing the |
| 132 | + // outcome, so a non-terminal request here is a broken invariant that |
| 133 | + // retrying cannot fix. |
| 134 | + metrics.NamedCounter(c.metricsScope, _opName, "invariant_errors", 1) |
| 135 | + return fmt.Errorf("request %s reached record in non-terminal state %q", request.ID, request.State) |
| 136 | + } |
| 137 | +} |
| 138 | + |
| 139 | +// advanceLastGreen points the queue's bookmark at request, retrying on version |
| 140 | +// conflicts. The bookmark only moves forward: a candidate whose id is not newer |
| 141 | +// than the stored one is skipped without a write, which also makes a redelivery |
| 142 | +// of the same request a no-op. |
| 143 | +// |
| 144 | +// Greenness comes from request.State rather than a persisted validation fact, |
| 145 | +// which is what record.md specifies. The two agree — a fact's degree is derived |
| 146 | +// from the same immutable state — and this reads the fact once the fact store |
| 147 | +// lands. |
| 148 | +func (c *Controller) advanceLastGreen(ctx context.Context, store storage.Storage, request entity.Request) error { |
| 149 | + queueStore := store.GetQueueStore() |
| 150 | + |
| 151 | + for { |
| 152 | + queueRow, err := queueStore.Get(ctx, request.Queue) |
| 153 | + if err != nil { |
| 154 | + return fmt.Errorf("failed to load queue %s to advance last green: %w", request.Queue, err) |
| 155 | + } |
| 156 | + |
| 157 | + newer, err := isNewerRequest(request.Queue, request.ID, queueRow.LastGreenRequestID) |
| 158 | + if err != nil { |
| 159 | + // Non-retryable: re-parsing the same ids cannot start succeeding. |
| 160 | + return err |
| 161 | + } |
| 162 | + if !newer { |
| 163 | + return nil |
| 164 | + } |
| 165 | + |
| 166 | + updated := queueRow |
| 167 | + updated.LastGreenURI = request.URI |
| 168 | + updated.LastGreenRequestID = request.ID |
| 169 | + newVersion := queueRow.Version + 1 |
| 170 | + if err := queueStore.Update(ctx, updated, queueRow.Version, newVersion); err != nil { |
| 171 | + if errors.Is(err, storage.ErrVersionMismatch) { |
| 172 | + continue |
| 173 | + } |
| 174 | + return fmt.Errorf("failed to advance last green for queue %s: %w", request.Queue, err) |
| 175 | + } |
| 176 | + |
| 177 | + metrics.NamedCounter(c.metricsScope, _opName, "last_green_advanced", 1) |
| 178 | + c.logger.Infow("advanced last green bookmark", |
| 179 | + "queue", request.Queue, |
| 180 | + "request_id", request.ID, |
| 181 | + "last_green_uri", request.URI, |
| 182 | + ) |
| 183 | + return nil |
| 184 | + } |
| 185 | +} |
| 186 | + |
| 187 | +// isNewerRequest reports whether candidate was ingested after current. An empty |
| 188 | +// current means the bookmark has never been set, so any candidate is newer. |
| 189 | +func isNewerRequest(queue, candidate, current string) (bool, error) { |
| 190 | + if current == "" { |
| 191 | + return true, nil |
| 192 | + } |
| 193 | + cmp, err := entity.CompareRequestID(queue, candidate, current) |
| 194 | + if err != nil { |
| 195 | + return false, fmt.Errorf("failed to compare request ids for queue %s: %w", queue, err) |
| 196 | + } |
| 197 | + return cmp > 0, nil |
| 198 | +} |
| 199 | + |
| 200 | +// loadRequest loads the request by id. |
| 201 | +func (c *Controller) loadRequest(ctx context.Context, store storage.Storage, id string) (entity.Request, error) { |
| 202 | + return loader.ByID(ctx, id, store.GetRequestStore().Get, "request") |
| 203 | +} |
| 204 | + |
| 205 | +// Name returns the controller name for logging and metrics. |
| 206 | +func (c *Controller) Name() string { |
| 207 | + return "record" |
| 208 | +} |
| 209 | + |
| 210 | +// TopicKey returns the topic key this controller subscribes to. |
| 211 | +func (c *Controller) TopicKey() consumer.TopicKey { |
| 212 | + return c.topicKey |
| 213 | +} |
| 214 | + |
| 215 | +// ConsumerGroup returns the consumer group for offset tracking. |
| 216 | +func (c *Controller) ConsumerGroup() string { |
| 217 | + return c.consumerGroup |
| 218 | +} |
0 commit comments