nocobase/nocobase · error · Error

job of head node (#${node.id}) not found in execution (#${th

Error message

job of head node (#${node.id}) not found in execution (#${this.execution.id})

What it means

This is the head-node variant of the rerun check: when rerun is called without nodeId but with overwrite=true, resolveRerun requires an existing job for the workflow's head node, because overwriting a rerun must replace an existing prior result. If the head node has no job recorded, it throws.

Source

Thrown at packages/plugins/@nocobase/plugin-workflow/src/server/Processor.ts:341

    try {
      await this.prepare();
      const node: FlowNodeModel = this.nodesMap.get(job.nodeId) as FlowNodeModel;
      await this.recall(node, job);
    } finally {
      this.leaveRunningState();
    }
  }

  private resolveRerun(options: ProcessorRerunOptions = {}) {
    const node = this.getRerunNode(options.nodeId);
    const targetJob = this.jobsMapByNodeKey[node.key];

    if (options.nodeId != null && !targetJob) {
      throw new Error(`job of node (#${node.id}) not found in execution (#${this.execution.id})`);
    }

    if (options.nodeId == null && options.overwrite && !targetJob) {
      throw new Error(`job of head node (#${node.id}) not found in execution (#${this.execution.id})`);
    }

    const input = this.getRerunInput(node);
    return { node, input, targetJob };
  }

  public async rerun(options: ProcessorRerunOptions = {}) {
    const { execution } = this;
    if (execution.status !== EXECUTION_STATUS.STARTED) {
      throw new Error(`execution (#${execution.id}) is not started`);
    }
    if (!(await this.shouldContinueExecution())) {
      this.logger.warn(`execution was ended with status ${execution.status} before, could not be rerun`, {
        workflowId: execution.workflowId,
      });
      return;
    }

View on GitHub (pinned to fa42722fef)

Solutions

Wait until the execution has at least run its head node before calling rerun with overwrite: true.
Call rerun() without overwrite (append mode) if you don't need to replace prior results.
If jobs were purged, re-trigger the workflow to create a new execution instead of rerunning.
Confirm the head node (node without upstream) matches the node that produced the job.

Example fix

// before
await processor.rerun({ overwrite: true }); // throws if nothing ran yet
// after
const head = execution.workflow.nodes.find((n) => !n.upstream);
const hasJob = await execution.countJobs({ where: { nodeId: head.id } });
await processor.rerun({ overwrite: hasJob > 0 });

Defensive patterns

Strategy: validation

Validate before calling

const head = workflow.nodes.find((n) => !n.upstream);
const headJob = head && (await execution.countJobs({ where: { nodeId: head.id } }));
if (overwrite && !headJob) {
  throw new Error('cannot overwrite-rerun: head node has not executed in this execution yet');
}

Type guard

function canOverwriteRerun(execution: ExecutionModel, workflow: WorkflowModel): boolean {
  const head = (workflow.nodes ?? []).find((n) => !n.upstream);
  return !!head && Array.isArray(execution.jobs) && execution.jobs.some((j) => j.nodeId === head.id);
}

Try / catch

try {
  await processor.rerun({ overwrite: true });
} catch (e) {
  if (/job of head node/.test(e.message)) {
    logger.warn('nothing executed yet; running append-mode rerun instead');
    await processor.rerun({});
    return;
  }
  throw e;
}

Prevention

Use overwrite: true only on executions with prior successful node runs.
Debounce/deduplicate rerun button clicks so a completed rerun isn't re-invoked.
Check execution status and job count before choosing overwrite mode.
Avoid editing the workflow head node while executions are pending rerun.

When it happens

Trigger: processor.rerun({ overwrite: true }) on an execution whose head node never produced a job — e.g. the execution started but failed before the head node executed, jobs were deleted, or the first node's key changed so jobsMapByNodeKey lookup misses.

Common situations: Overwrite-rerun of a freshly created execution that has not executed any node yet; rerun after job cleanup jobs purged the head job; workflow edited so the current head node differs from the one that ran.

Related errors

AI-assisted analysis of nocobase/nocobase@fa42722fef (2026-09-01). Data as JSON: /api/errors/525a72aebd94a21b. Report an issue: GitHub.