Fix: n8n failed with HTTP 500

    An HTTP 500 means the PostFast request reached the server but failed there. Identify the failing operation, validate the payload, and retry only after checking whether the post was created.

    Last updated 2026-09-21~9 min read
    n8nPostFast APIGoogle SheetsOpenAI

    Last updated: 2026-06-08 · Tested on: n8n 1.x · Read time: 5 min

    Symptoms — what you're seeing

    • An n8n node returns an HTTP 500 or “Internal Server Error” when it creates or updates a scheduled social post
    • The failed request reaches the PostFast API, but the response body contains a server error instead of a usable post ID
    • A Google Sheets content calendar creates duplicate posts after a 500 response or request timeout and n8n retries
    • Posts work for one Facebook Page or LinkedIn organization but return 500 when the brand has multiple connected accounts
    • Instagram, TikTok, YouTube, or Pinterest requests return 500 when required media is missing, invalid, or too large

    This page covers an HTTP 500 returned by the PostFast request in an n8n workflow. It is different from a post that was created successfully and failed later in an asynchronous publishing job. First identify the node and request that returned 500; then validate the payload and check whether the server created a post before retrying.

    Why this happens

    1. The 500 is coming from a specific create or upload request. An n8n execution can contain successful Google Sheets and OpenAI steps before the PostFast node fails. Open the failed node and inspect its HTTP status and response body. A 500 from the create request needs a payload and server-response investigation; a later failed post status needs a publishing-status investigation.

    2. The workflow does not select one connected account. A brand can have two Facebook Pages or two LinkedIn organizations connected to the same system. A label such as “LinkedIn” is not enough. Each post needs one account ID, and that account must have connectionStatus set to CONNECTED. An invalid or disconnected destination can surface as a server-side 500 instead of a useful validation error.

    3. Required media is not validated before create. TikTok, YouTube, Instagram, and Pinterest require media even when the post is a draft. The media type must also match the uploaded file. File-size limits are enforced during upload, including the documented 250 MB general video cap, 100 MB Bluesky cap, and 50 MB Telegram cap. Put validation before the create request so an invalid payload does not reach the server.

    4. A 500 or timeout is being treated as proof that no post was created. The create request has no idempotency key. The server can create the post and fail while returning the response, or the response can be lost after creation. Blindly retrying can create a duplicate scheduled post. Search for the existing post before sending another create request.

    5. The workflow waits for an asynchronous operation. The create call schedules the post. Publishing runs later as a server-side job. A Wait node immediately after create only burns an execution while it waits; it does not turn an HTTP 500 response into a publish confirmation or explain a later post-status failure.

    6. Approval is a state, not an execution step. A post created with approval pending will not publish until someone approves it in the application or through an agent. The n8n workflow needs a branch for that state: send pending posts to the approval channel, then handle approved posts separately. Do not retry a pending post as if it were an HTTP 500.

    7. Bulk requests exceed limits or expose one bad row. The create endpoint allows 150 creates per minute and 350 per day. The API key also has limits of 60 requests per minute, 150 per five minutes, 300 per hour, and 2,000 per day. A 500-row Google Sheets result sent directly into a create node can produce rate-limit responses, server strain, or a failing row that is difficult to isolate.

    8. Link posts can hit a separate platform-plan limit. X link posts have monthly limits that vary by plan. If an OpenAI step appends a link to every generated post, the workflow can exhaust that allowance even when the n8n execution itself is successful.

    Stuck on the diagnosis? Book a 30-min call. I'll inspect the failed request, account selection, and retry path with you. You can also email Basit.

    The fix, step by step

    1. Confirm that the failure is an HTTP 500. Open the failed n8n node and record the node name, operation, HTTP status, response body, request URL, and execution time. Confirm that the 500 came from the PostFast create, upload, or update request rather than from a later status check. If the response body contains a request or correlation ID, keep it with the execution record — the exact fields returned vary by PostFast node/API version, so check the response payload directly rather than assuming a fixed shape.
    2. Resolve the destination account first. Add the PostFast Get Many Social Accounts operation before the create operation. Filter the results to the account you intend to use, then pass exactly one account ID into the post payload. Add an IF node that stops the item unless the selected account has connectionStatus equal to CONNECTED. Do not let the workflow select the first account returned by the API.
    3. Validate every item before creating a post. For a workflow receiving multiple rows from Google Sheets, use a Code node that validates each item independently. This keeps valid rows from being silently dropped because the first row failed.
      const items = $input.all();
      
      return items.map((item, index) => {
        const post = item.json;
        const platform = post.platform;
        const hasMedia = Array.isArray(post.media) && post.media.length > 0;
      
        // Field names should match your PostFast node/API version — adjust if needed
        const mediaRequired = ["TikTok", "YouTube", "Instagram", "Pinterest"].includes(platform);
      
        return {
          json: {
            ...post,
            validation: {
              row: index + 1,
              accountPresent: Boolean(post.accountId),
              mediaPresent: hasMedia,
              mediaRequired,
              valid: Boolean(post.accountId) && (!mediaRequired || hasMedia),
            },
          },
        };
      });
      After the Code node, use an IF node on validation.valid. Route invalid items to an error branch that records the row, platform, and reason. Route only valid items to the PostFast create operation. Validate the media file type and size before upload as well; presence alone is not enough.
    4. Test one known-good item. Send one item with a confirmed connected account, a valid scheduled time, plain text, and media only when the platform requires it. Compare this request with the failing request field by field. If the minimal item succeeds, add the original fields back one at a time until the field that causes the 500 is isolated. Note that not every PostFast platform accepts a text-only test post, so pick a platform that does before assuming the payload is the problem.
    5. Store the returned post ID. After a successful create response, write the returned post ID alongside the source row ID, account ID, scheduled time, and workflow execution ID. A Google Sheets Update Row node or a Supabase table works for this state. The important part is that a later run can determine whether the row already produced a scheduled post.
    6. Turn off blind retries on create. Do not enable Retry On Fail for the create request while the endpoint has no idempotency key. If the request returns 500 or times out, run Get Many Social Posts filtered by the selected account and scheduled time before creating anything again. If a matching post exists, store its ID and continue. If no matching post exists, create once and record the response. This is a check-before-retry pattern, not an idempotency guarantee, so keep the search window narrow enough to avoid matching an unrelated post.
    7. Do not poll for publishing with an immediate Wait node. Treat a successful create as “scheduled” and finish the workflow after recording the post ID. If confirmation matters, run a separate scheduled workflow later that calls Get Many Social Posts and checks the post status. A failed status belongs in an alert branch; it does not mean the original n8n create execution should remain open.
    8. Branch on approval status. Create the post with the intended approval state. If the response is pending, send the post ID, account, scheduled time, and content preview to Slack or the approval system. Do not treat a pending post as a failed post. When the approval action happens, update or approve the existing post rather than creating a second one.
    9. Batch large inputs. Use Loop Over Items or a controlled batch pattern to process a limited number of rows, then add a Wait node between batches. Keep the batch size and delay below both the create endpoint limits and the API-key limits. Count all API calls in the batch, including account lookups, media uploads, post creates, and status checks. A workflow that only counts create nodes can still exceed the key-level request limit.
    10. Control generated links. If an OpenAI node generates the post copy, make link inclusion an explicit field in the input instead of appending a URL to every item. Track link posts by platform and month before the create step. If the workflow has reached the applicable X plan limit, route the item to an approval or rewrite branch instead of submitting another link post.

    Inline checkpoint: If the failed node is a Schedule Trigger rather than the PostFast request, or if the post is created for the wrong time instead of returning HTTP 500, check the n8n schedule trigger timezone diagnostic before changing retry logic. A wrong schedule is a timezone issue, not a 500 response.

    Common variations

    • “The create node returns 500 immediately.” Inspect the request and response body, then test one known-good item. Check the account ID, connection status, scheduled time, required media, and payload field names before retrying.
    • “The workflow creates posts but they fail later.” This is not the same as an HTTP 500 from create. Check the selected account's connection status, required media, media type, and the later post status.
    • “The same post appears twice after a 500 or timeout.” Disable automatic retries on create and search Get Many Social Posts before retrying. The create endpoint does not currently provide an idempotency key.
    • “Only carousel or video posts return 500.” Check that the media array is present and that every file matches the platform's accepted type and size. Image-only tests do not validate a video or slideshow path.
    • “A weekly OpenAI workflow starts returning 500 or rate-limit errors.” Inspect the failing node first, count requests across the whole workflow, batch the rows, and add a Wait node between chunks. The OpenAI step can also have its own rate limit; see the n8n OpenAI 429 diagnostic separately.
    • “The post is waiting forever.” Check whether it was created with approval pending. Approval is a separate state and requires an explicit approval action; it is not evidence of an HTTP 500.
    • “Every generated post contains a link.” Make link inclusion a controlled input. X link-post allowances are plan-specific, so appending a URL in an n8n expression can exhaust the monthly limit without producing a node error.

    When this isn't the issue

    If the failing node did not return HTTP 500, or if the request returned a post ID successfully, check the next diagnostic forks:

    • Inspect the post returned by Get Many Social Posts and distinguish pending, approved, failed, and published states — the exact status values can vary by PostFast API version, so confirm them against the actual response.
    • Check whether the workflow's schedule is shifted by the n8n instance timezone. The schedule trigger timezone fix covers the UTC-versus-local-time failure separately.
    • Check the raw HTTP response for authentication or permission errors. A 401 or 403 is not a 500; if the failing request is to a different API rather than PostFast, use the relevant n8n HTTP 401 diagnostic instead of changing the social-post retry pattern.

    Bottom line: confirm the failing request is an HTTP 500, isolate it with one known-good item, resolve one connected account, validate every item before create, store the returned post ID, and check for an existing post before retrying. Do not keep a Wait node open waiting for a server-side publish job.

    Related pages

    Stuck on this for more than 30 minutes? Book a 30-min call. I'll get on Zoom, look at your workflow, and either fix it on the call or scope what it'll take. No upsell, no decks. You can also email Basit.

    Need this workflow customized or debugged? Book a 30-min scope call →

    Stuck configuring this workflow?

    I build and fix n8n workflows for agencies and SaaS teams. 30-minute scope call — bring the problem, leave with an estimate.

    Related pages