Files
AgentCompany/src/main/orchestration/run-dispatch.ts
T
ZacharyZhang-NY 2dd62e5338 Adapt heartbeat/task-runner/turn-context to lockMode=none first-class path
- heartbeat.ts and task-runner.ts now share a single ResolveWorkspaceFn
  signature exported from run-dispatch.ts. Callers handle
  WorkspaceConfigurationError explicitly:
    * heartbeat: release the task lease and skip with new
      workspace_unresolved reason (task re-enters pool next heartbeat)
    * task-runner: release the lease and return WORKSPACE_UNRESOLVED
      to the caller
  No silent fallback to a random shared workspace.
- crash-recovery.ts imports WorkspaceConfigurationError for clean
  catch-branch attribution; the outer try already rolls back the
  queued run on any resolve failure.
- run-dispatch.ts exposes ResolveWorkspaceFn so downstream modules
  type their injected resolver against the same shape (single source
  of truth).
- HeartbeatSkipReason gains workspace_unresolved to keep the closed
  enum honest about all heartbeat exit paths.
2026-04-20 15:32:55 -04:00

132 lines
4.9 KiB
TypeScript

import { getConnectorExecutionReadinessIssue } from "@shared/connector-policy";
import type { AgentRecord, ConnectorRecord, DepartmentId, TaskRecord } from "@shared/types";
import type { OrchestrationDependencies } from "./types";
import { buildRunPayload, type RunDispatchParams } from "./turn-context";
import { validateConnectorModelSelection } from "../connector-model-policy";
import { resolveWorkspaceWithLockMode } from "./resolve-workspace";
type ResolveWorkspaceTaskInput =
| (Pick<TaskRecord, "workspaceId" | "projectId"> & { requiresFilesystem?: boolean })
| null;
type ResolveWorkspaceAgentInput = Pick<AgentRecord, "workspaceId"> & {
role?: string | null;
department?: DepartmentId | null;
};
export type ResolveWorkspaceFn = (
task: ResolveWorkspaceTaskInput,
agent: ResolveWorkspaceAgentInput,
companyId: string,
) => ReturnType<typeof resolveWorkspaceWithLockMode>;
export function createRunDispatch(deps: OrchestrationDependencies) {
const {
db,
logger,
isQuitting,
hasWorkerReady,
getApiPort,
createRunToken,
enqueueWorkerRun,
listRuntimeSkillEntries,
budgetService,
getSessionHandoff,
evaluatePreRunSessionRotation,
} = deps;
const resolveWorkspace: ResolveWorkspaceFn = (task, agent, companyId) =>
resolveWorkspaceWithLockMode(db, task, agent, companyId);
function getRunEligibilityFailure(
agent: { status: string; budgetMonthlyUsd: number; spentMonthlyUsd: number; name: string; connectorId: string; id: string; companyId: string; model?: string | null },
connector: Pick<ConnectorRecord, "id" | "label" | "status" | "capabilityMatrix">,
projectId: string | null = null,
) {
if (["terminated", "paused", "pending_approval"].includes(agent.status)) {
return "AGENT_NOT_RUNNABLE";
}
const gate = budgetService.evaluateBudgetGate(agent.companyId, agent.id, projectId);
if (!gate.allowed) {
logger.info(`[run] Agent ${agent.name} budget policy blocked: ${gate.blockedBy.map(b => `${b.scopeType}(${b.policyId})`).join(", ")}`);
return "BUDGET_POLICY_BLOCKED";
}
const connectorIssue = getConnectorExecutionReadinessIssue(connector);
if (connectorIssue) {
const issue = connectorIssue ?? `Connector ${connector.id} is not execution-ready.`;
logger.info(`[run] Agent ${agent.name} connector ${connector.id} blocked: ${issue}`);
return "CONNECTOR_NOT_READY";
}
const modelSelection = validateConnectorModelSelection(
{
getConnector: (id) => db.connectorsCrud.getConnector(id),
getDisabledModels: (cid, connId) => db.settings.getDisabledModels(cid, connId),
},
agent.companyId,
connector.id,
agent.model ?? null,
);
if (!modelSelection.ok) {
logger.info(`[run] Agent ${agent.name} connector model blocked: ${modelSelection.error}`);
return "CONNECTOR_MODEL_NOT_ALLOWED";
}
return null;
}
function dispatchRunToWorker(params: RunDispatchParams) {
if (isQuitting()) {
throw new Error("Application is shutting down — refusing to dispatch new run.");
}
if (!hasWorkerReady()) {
throw new Error("Orchestration worker is not ready.");
}
const connectorIssue = getConnectorExecutionReadinessIssue(params.connector);
if (connectorIssue) {
throw new Error(connectorIssue);
}
// D-02: Claim the workspace lock synchronously before enqueuing the run.
// Previously the worker claimed the lock only after it emitted `running`,
// so two runs could race between enqueue and startup and both read the
// same workspace mid-write. Main is the authoritative coordinator.
//
// On contention we mark the run `cancelled`, not `failed`. The Dashboard
// "Needs attention" card only surfaces failed/timed_out/interrupted runs
// (see src/renderer/components/dashboard/Dashboard.tsx), so the user is
// not alarmed by a transient backoff. The task row is rolled back to
// `todo` via the `finishRun` never-started branch, so heartbeat picks
// it up on the next cycle with no manual intervention.
if (
params.workspaceMode === "exclusive" &&
params.workspace &&
!db.workspacesMod.claimWorkspaceLock(params.workspace.id, params.runId)
) {
db.runs.finishRun({
runId: params.runId,
status: "cancelled",
summary: null,
errorMessage: `Workspace ${params.workspace.id} is in use by another run; will retry on the next heartbeat.`,
exitCode: null,
signal: null,
model: null,
unattributedCost: true,
});
return;
}
const payload = buildRunPayload(params, {
db,
logger,
getApiPort,
createRunToken,
listRuntimeSkillEntries,
getSessionHandoff,
evaluatePreRunSessionRotation,
});
enqueueWorkerRun(payload);
}
return { resolveWorkspace, getRunEligibilityFailure, dispatchRunToWorker };
}