{
  "id": "retryOneHttpItem",
  "name": "Retry One HTTP Item at a Time",
  "description": "Process two synthetic HTTP operations sequentially, with a bounded retry for each input item and a combined done output.",
  "active": false,
  "nodes": [
    {
      "id": "start",
      "name": "Start",
      "type": "n8n-nodes-base.manualTrigger",
      "typeVersion": 1,
      "position": [
        -400,
        0
      ],
      "parameters": {}
    },
    {
      "id": "fixture",
      "name": "Fixture",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [
        -200,
        0
      ],
      "parameters": {
        "jsCode": "return [\n  {\n    \"shortId\": 42,\n    \"styling\": {\n      \"color\": \"#112233\",\n      \"fontSize\": 31\n    }\n  },\n  {\n    \"shortId\": 43,\n    \"styling\": {\n      \"color\": \"#aabbcc\",\n      \"fontSize\": 47\n    }\n  }\n].map(json => ({ json }));"
      },
      "notes": "Two synthetic inputs with distinct IDs and styling. Replace with your source items after validating the test service."
    },
    {
      "id": "render",
      "name": "Render",
      "type": "n8n-nodes-base.httpRequest",
      "typeVersion": 4.2,
      "position": [
        220,
        180
      ],
      "parameters": {
        "method": "POST",
        "url": "=https://api.example.com/renders/{{ $json.shortId }}",
        "sendBody": true,
        "specifyBody": "json",
        "jsonBody": "={{ { shortId: $json.shortId, renderOptions: $json.styling } }}",
        "options": {
          "timeout": 10000
        }
      },
      "retryOnFail": true,
      "maxTries": 2,
      "waitBetweenTries": 1000,
      "notes": "Replace the api.example.com placeholder with your test endpoint. Each node execution receives one item; two total attempts are allowed."
    },
    {
      "id": "done",
      "name": "Done",
      "type": "n8n-nodes-base.noOp",
      "typeVersion": 1,
      "position": [
        220,
        -100
      ],
      "parameters": {}
    },
    {
      "id": "loop",
      "name": "Loop Over Items",
      "type": "n8n-nodes-base.splitInBatches",
      "typeVersion": 3,
      "position": [
        0,
        0
      ],
      "parameters": {
        "batchSize": 1,
        "options": {}
      }
    },
    {
      "id": "setup-note",
      "name": "Import and configure",
      "type": "n8n-nodes-base.stickyNote",
      "typeVersion": 1,
      "position": [
        -590,
        -720
      ],
      "parameters": {
        "content": "## Retry one HTTP item at a time\n**Loop Over Items → Batch Size = 1** gives each HTTP node execution one input item. The retry budget applies to that item.\n\n### Run in the editor\n1. Import this JSON with **Import from File**.\n2. Open **Render** and replace `https://api.example.com` with your test service base URL, keeping `/renders/{{ $json.shortId }}`. Configure your test service credentials in that node as needed.\n3. Click **Execute workflow**. **Fixture** supplies two synthetic operations, 42 and 43, each with its own nested styling object.\n4. Inspect **Done** for the two completed HTTP response items.\n\nThe loop output goes to Render; Render connects back to Loop Over Items; the done output goes to Done. Keep **Reset** off for this fixed input list.\n\n**Tested:** n8n 2.41.7, HTTP Request v4.2, Loop Over Items v3; 2026-10-06.",
        "width": 715,
        "height": 600
      }
    },
    {
      "id": "checks-note",
      "name": "Expected requests and adaptation",
      "type": "n8n-nodes-base.stickyNote",
      "typeVersion": 1,
      "position": [
        160,
        -720
      ],
      "parameters": {
        "content": "## Verify the retry boundary\nFor a reproducible test, configure your test service as follows:\n- `POST /renders/42`: first response **503**, then **200** with `{\"id\":\"render-42\"}`.\n- `POST /renders/43`: always **200** with `{\"id\":\"render-43\"}`.\n\n**Observed:** 42→503, 42→200, 43→200. Done receives `render-42` and `render-43`. Both request attempts for 42 retain its complete JSON body.\n\nRender uses **Max Tries = 2**, **Wait Between Tries = 1000 ms**, and a **10-second timeout**. A second failure stops this workflow. HTTP Request's built-in batching paces items inside a node execution; the outer loop defines this per-item retry boundary.\n\nFor real writes, reuse a stable business operation key across retries and use the destination's durable idempotency contract to recover an accepted write whose response is lost.\n\nOptional [n8n-check fixture and request traces](https://github.com/blucca/n8n-check/tree/v0.1.1/examples/retry-isolation) provide a local mock and JSON/JUnit checks for this pattern. The editor workflow uses standard n8n nodes.",
        "width": 735,
        "height": 600
      }
    }
  ],
  "connections": {
    "Start": {
      "main": [
        [
          {
            "node": "Fixture",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Fixture": {
      "main": [
        [
          {
            "node": "Loop Over Items",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Render": {
      "main": [
        [
          {
            "node": "Loop Over Items",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Loop Over Items": {
      "main": [
        [
          {
            "node": "Done",
            "type": "main",
            "index": 0
          }
        ],
        [
          {
            "node": "Render",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  },
  "settings": {
    "executionOrder": "v1"
  }
}