Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 | 15x 23x 23x 23x 23x 23x 23x 23x 23x 23x | /**
* AgentLifecycleController — Unified orchestrator for complete agent lifecycle.
*
* Replaces scattered lifecycle logic with comprehensive orchestration including:
* - Always clear signal.json before spawn/resume
* - Robust process completion waiting
* - Retry up to 3 times with comprehensive error handling
* - Auth/usage limit error detection with account switching
* - Missing signal recovery with instruction prompts
* - Debug mode archival vs production cleanup
*/
import { createModuleLogger } from '../../logger/index.js';
import type { AgentRepository } from '../../db/repositories/agent-repository.js';
import type { AccountRepository } from '../../db/repositories/account-repository.js';
import type { ProcessManager } from '../process-manager.js';
import type { CleanupManager } from '../cleanup-manager.js';
import type { SpawnAgentOptions } from '../types.js';
import type { SignalManager, SignalData } from './signal-manager.js';
import type { RetryPolicy, AgentError } from './retry-policy.js';
import { AgentExhaustedError, AgentFailureError } from './retry-policy.js';
import type { AgentErrorAnalyzer } from './error-analyzer.js';
import type { CleanupStrategy, AgentInfo } from './cleanup-strategy.js';
const log = createModuleLogger('lifecycle-controller');
export interface CompletionResult {
success: boolean;
signal?: SignalData;
error?: Error;
exitCode?: number | null;
stderr?: string;
}
export interface ResumeAgentOptions {
agentId: string;
answers: Record<string, string>;
}
export class AgentLifecycleController {
constructor(
private signalManager: SignalManager,
private retryPolicy: RetryPolicy,
private errorAnalyzer: AgentErrorAnalyzer,
private processManager: ProcessManager,
private repository: AgentRepository,
private cleanupManager: CleanupManager,
private cleanupStrategy: CleanupStrategy,
private accountRepository?: AccountRepository,
private debug: boolean = false,
) {}
/**
* Execute spawn operation with comprehensive retry and error handling.
* Always clears signal.json before starting and waits for process completion.
*/
async spawnWithRetry(
spawnFn: (options: SpawnAgentOptions) => Promise<AgentInfo>,
options: SpawnAgentOptions
): Promise<AgentInfo> {
log.info({
taskId: options.taskId,
provider: options.provider,
initiativeId: options.initiativeId,
mode: options.mode
}, 'starting agent spawn with retry');
return this.executeWithRetry('spawn', spawnFn, options);
}
/**
* Execute resume operation with comprehensive retry and error handling.
* Always clears signal.json before resuming and waits for process completion.
*/
async resumeWithRetry(
resumeFn: (agentId: string, answers: Record<string, string>) => Promise<void>,
options: ResumeAgentOptions
): Promise<void> {
log.info({
agentId: options.agentId,
answerKeys: Object.keys(options.answers)
}, 'starting agent resume with retry');
await this.executeWithRetry('resume', async () => {
await resumeFn(options.agentId, options.answers);
const agent = await this.repository.findById(options.agentId);
if (!agent) throw new Error(`Agent '${options.agentId}' not found after resume`);
return this.toAgentInfo(agent);
}, options);
}
/**
* Main retry orchestrator for spawn/resume operations.
*/
private async executeWithRetry<T>(
operation: 'spawn' | 'resume',
operationFn: (options: T) => Promise<AgentInfo>,
options: T
): Promise<AgentInfo> {
for (let attempt = 1; attempt <= this.retryPolicy.maxAttempts; attempt++) {
try {
log.debug({ operation, attempt, maxAttempts: this.retryPolicy.maxAttempts }, 'starting attempt');
// Execute operation
const agent = await operationFn(options);
const agentWorkdir = this.processManager.getAgentWorkdir(agent.worktreeId);
// CRITICAL: Always clear signal.json before start
log.debug({ agentId: agent.id, agentWorkdir }, 'clearing signal.json before process start');
await this.signalManager.clearSignal(agentWorkdir);
// Wait for process completion with robust detection
const result = await this.waitForCompletion(agent);
if (result.success) {
// Handle post-completion cleanup
await this.handlePostCompletion(agent);
log.info({
agentId: agent.id,
name: agent.name,
attempt,
operation
}, 'agent lifecycle completed successfully');
return agent;
}
// Analyze error and determine retry strategy
const agentError = await this.errorAnalyzer.analyzeError(
result.error || new Error('Unknown completion failure'),
result.exitCode,
result.stderr,
agentWorkdir
);
// Persist error to DB if required
if (agentError.shouldPersistToDB) {
await this.persistError(agent.id, agentError);
}
// Handle account switching for usage limits
if (agentError.requiresAccountSwitch) {
await this.handleAccountExhaustion(agent.id);
throw new AgentExhaustedError(agentError.message, agentError);
}
// Check if should retry
if (!this.retryPolicy.shouldRetry(agentError, attempt)) {
log.warn({
agentId: agent.id,
errorType: agentError.type,
attempt,
maxAttempts: this.retryPolicy.maxAttempts
}, 'max retry attempts reached or error not retriable');
throw new AgentFailureError(agentError.message, agentError);
}
// Handle special retry cases
if (agentError.type === 'missing_signal') {
// This would need to modify the options to add instruction prompt
// For now, log the special case
log.info({
agentId: agent.id,
attempt
}, 'will retry with missing signal instruction (not yet implemented)');
}
// Wait before retry
const delay = this.retryPolicy.getRetryDelay(attempt);
log.info({
agentId: agent.id,
attempt,
delay,
errorType: agentError.type,
errorMessage: agentError.message
}, 'retrying after delay');
await this.delay(delay);
} catch (error) {
if (error instanceof AgentExhaustedError || error instanceof AgentFailureError) {
throw error; // Don't retry these
}
if (attempt === this.retryPolicy.maxAttempts) {
log.error({
operation,
attempt,
error: error instanceof Error ? error.message : String(error)
}, 'final attempt failed, giving up');
throw error;
}
log.warn({
operation,
attempt,
error: error instanceof Error ? error.message : String(error)
}, 'attempt failed, will retry');
}
}
throw new Error('Unexpected: retry loop completed without success or terminal error');
}
/**
* Wait for process completion with robust signal detection.
* Replaces scattered completion detection with unified approach.
*/
private async waitForCompletion(agent: AgentInfo): Promise<CompletionResult> {
const agentWorkdir = this.processManager.getAgentWorkdir(agent.worktreeId);
log.debug({
agentId: agent.id,
name: agent.name,
agentWorkdir
}, 'waiting for process completion');
// Wait for process to exit (this would need integration with ProcessManager)
// For now, simulate with a timeout approach
// TODO: Implement waitForProcessCompletion in ProcessManager
// Wait for signal within reasonable timeout (30 seconds)
const signal = await this.signalManager.waitForSignal(agentWorkdir, 30000);
if (signal) {
log.debug({
agentId: agent.id,
signalStatus: signal.status
}, 'agent completed with valid signal');
return { success: true, signal };
}
// No signal found - this is an error condition
log.warn({
agentId: agent.id,
agentWorkdir
}, 'process completed without valid signal.json');
return {
success: false,
error: new Error('Process completed without valid signal.json'),
exitCode: null // Would get from ProcessManager
};
}
/**
* Handle post-completion cleanup based on agent status and debug mode.
*/
private async handlePostCompletion(agent: AgentInfo): Promise<void> {
// Only cleanup if agent is not waiting for user input
if (agent.status === 'waiting_for_input') {
log.debug({ agentId: agent.id }, 'agent waiting for input, skipping cleanup');
return;
}
try {
const cleanupAction = await this.cleanupStrategy.shouldCleanup(agent, this.debug);
await this.cleanupStrategy.executeCleanup(agent, cleanupAction);
log.debug({
agentId: agent.id,
name: agent.name,
cleanupAction
}, 'post-completion cleanup executed');
} catch (error) {
log.warn({
agentId: agent.id,
error: error instanceof Error ? error.message : String(error)
}, 'post-completion cleanup failed');
}
}
/**
* Persist error details to database for debugging.
*/
private async persistError(agentId: string, error: AgentError): Promise<void> {
try {
const errorData = {
errorType: error.type,
errorMessage: error.message,
exitCode: error.exitCode,
isTransient: error.isTransient,
requiresAccountSwitch: error.requiresAccountSwitch,
updatedAt: new Date(),
};
// This would need database schema updates to store error details
// For now, just update with basic error info
await this.repository.update(agentId, {
exitCode: error.exitCode,
updatedAt: new Date(),
});
log.debug({
agentId,
errorType: error.type,
exitCode: error.exitCode
}, 'error details persisted to database');
} catch (dbError) {
log.warn({
agentId,
error: dbError instanceof Error ? dbError.message : String(dbError)
}, 'failed to persist error to database');
}
}
/**
* Handle account exhaustion by marking account as exhausted.
*/
private async handleAccountExhaustion(agentId: string): Promise<void> {
if (!this.accountRepository) {
log.debug({ agentId }, 'no account repository available for exhaustion handling');
return;
}
try {
const agent = await this.repository.findById(agentId);
if (!agent?.accountId) {
log.debug({ agentId }, 'agent has no account ID for exhaustion handling');
return;
}
// Mark account as exhausted for 1 hour
const exhaustedUntil = new Date(Date.now() + 60 * 60 * 1000);
await this.accountRepository.markExhausted(agent.accountId, exhaustedUntil);
log.info({
agentId,
accountId: agent.accountId,
exhaustedUntil
}, 'marked account as exhausted due to usage limits');
} catch (error) {
log.warn({
agentId,
error: error instanceof Error ? error.message : String(error)
}, 'failed to mark account as exhausted');
}
}
/**
* Simple delay utility for retry backoff.
*/
private delay(ms: number): Promise<void> {
return new Promise(resolve => setTimeout(resolve, ms));
}
/**
* Convert database agent record to AgentInfo.
*/
private toAgentInfo(agent: any): AgentInfo {
return {
id: agent.id,
name: agent.name,
status: agent.status,
initiativeId: agent.initiativeId,
worktreeId: agent.worktreeId,
};
}
} |