Guide · retries and reconnects
Error handling across Make, n8n, and Zapier
What a green run means — and who retries or reconnects. Totals are not published yet. Self-hosted is a separate lane. A score is not a green light. This page does not add new scores.
Handle this job, or don’t over-trust a green run
Handle
Handle / dig in if…
A small-team job is live or about to ship — form to CRM, invoice to sheet plus Slack, or a shared inbox — and you need to know what each product calls success. Who retries a rate-limit miss. Who can click OAuth reconnect when the contractor’s Gmail expires. Who is actually emailed. Confirm the last destination write before you trust a green icon.
Defer
Defer or don’t over-trust green runs if…
You still need to name the kind of failure — that is When automations fail. The switch still looks on and you have not opened the last write — When automations look fine. You are counting tasks, credits, or executions — How to meter automation. Nobody is named to reconnect — Who should own your automations. This page compares retries and reconnect ownership. It does not rank products.
A handle-or-don’t-over-trust-green-runs check for one live job.
- You are picking or running Make vs n8n vs Zapier and need to know what a green run actually means.
- You need who retries, who owns the OAuth reconnect, and who gets emailed.
Read more
What this page tells you. How small teams should think about errors, retries, and reconnects when picking — or operating — Make vs n8n vs Zapier. What each UI calls success, who retries, who owns the OAuth reconnect, who gets notified, and how a false-green run differs from a hard fail. Classes stay on the failure taxonomy. The operator checklist stays on When automations look fine.
GuideSome scores not shown yetVendor and community pages accessed 2026-09-17Live scorecard
This page vs the taxonomy and look-fine
When automations fail — a practical taxonomy is the class list: auth, bad data, rate limits, timeouts, quota, handled versus unhandled, and expected non-match. This page does not reprint that table.
When automations look fine is the operator story — reconnect loops, false-green runs, who gets paged, what to log. This page does not reprint that checklist.
Units and cost live on How to meter automation and Same job, three meters. This page does not invent a monthly dollar. It is the product compare: what “success” means in each UI, how retries and handlers work, who owns the reconnect, and who is actually emailed. Then name the owner on Who should own your automations.
Compare lanes
Evidence from primary vendor docs and community threads. Access date: 2026-09-17. Totals are not published yet. This page does not add a score.
Not scored yet
Make
Make’s overview of error handling names five handlers. Skip and Resume can finish as Success after a module error — Skip drops the bundle; Resume writes a substitute. Retry stores an incomplete execution and ends as Warning. Commit stops and keeps committed changes (Warning). Rollback — the default when you set no handler and leave incomplete executions off — stops the run as Error.
Incomplete executions are off by default. Turn them on to store unfinished runs. ConnectionError, RateLimitError, and ModuleTimeoutError retry automatically once that setting is on. A Make Community 429 thread is wait-and-retry, not a rebuild. Default consecutive errors: 3, then scheduling stops. AccountValidationError, OperationsLimitExceededError, and DataSizeLimitExceededError disable scheduling at once and cannot wait out the consecutive-error count.
Honest cases: When Make is the right pick.
Source: Make — overview of error handling · Make — retry error handler · Make — incomplete executions · Make — automatic retry of incomplete executions · Make — fix errors and warnings · Make Community — 429 rate limit / wait and try again · accessed 2026-09-17
Not scored yet
n8n Cloud
The Executions list filters Failed, Running, Success, or Waiting. Retry a failed execution from that list — with the currently saved workflow or the original. There is no account-wide Autoreplay. Set an error workflow in Workflow Settings. It must start with the Error Trigger. Use Stop And Error when you want a run to fail on purpose and fire that workflow.
The Error Trigger runs on automatic failures, not on a manual test. n8n does not mail the owner unless you wire email or Slack on that error workflow. A saved OAuth credential can still sit in the list after the token stops refreshing — community threads describe a reconnect click as the recovery. Same editor on self-host; the box is a different job.
Honest cases: When n8n self-host is worth it if you will run the box. Cloud pick stays on Which automation tool.
Source: n8n — handle errors gracefully · n8n — Error Trigger · n8n — Stop And Error · n8n — view all executions · accessed 2026-09-17
Not scored yet
Zapier
Run history names more than a green or red pair. The statuses that matter for this page: Successful, Errored, Handled error, On hold, Safely halted, and Scheduled (Autoreplay). Successful is a completed path. Safely halted is usually a search with no row — a purposeful stop, not a success write. Handled error means a custom error handler ran an alternative workflow. The Zap will not auto-off on handled errors.
Autoreplay retries Errored steps up to five times. The run sits Scheduled until the last attempt. Zapier does not email until that last attempt fails. Custom error handlers (Team and Enterprise) turn Autoreplay off for that Zap and send no error email. Autoreplay does not replay On hold or Safely halted. Reconnect is only for connections you own.
Honest cases: When Zapier is still the right pick.
Source: Zapier — review run statuses in Zap workflows · Zapier — manage notifications when errors occur · Zapier — set up custom error handling · Zapier — what is replay · Zapier — replay Zap runs · accessed 2026-09-17
Five questions, three products
Same five jobs on each lane. Not a 1–5 score. Confirm the destination write before you trust a green icon.
Swipe sideways on a narrow screen. Not a scored table.
| Question | Make | n8n Cloud | Zapier |
|---|---|---|---|
| What “success” means | Skip and Resume can mark the run Success even after a module error. Retry and Commit end as Warning. Rollback — the default with incomplete executions off — ends as Error. | The Executions list filters Failed, Running, Success, or Waiting. A Success filter is the completed run, not a destination write you have checked. | Successful is a completed path. Handled error means an alternative workflow ran. Safely halted is a purposeful stop, usually a search with no row — not a success write. |
| Retries and error handlers | Incomplete executions are off by default. Enable them to store unfinished runs. Retry stores a bundle and can auto-complete. ConnectionError, RateLimitError, and ModuleTimeoutError retry automatically once incomplete executions are on. Default consecutive errors: 3, then scheduling stops. | No vendor-wide Autoreplay. Retry a failed execution from the Executions list — with the currently saved workflow or the original. Set an error workflow in Workflow Settings; it must start with the Error Trigger. | Autoreplay retries Errored steps up to five times and holds the run as Scheduled. Custom error handlers (Team and Enterprise) turn Autoreplay off for that Zap. Autoreplay does not replay On hold or Safely halted. |
| OAuth reconnect ownership | OAuth access is time-limited. Reauthorize, or create a new connection. AccountValidationError (HTTP 401/403) happens during setup and cannot be swallowed by an error handler. A community thread names a Google app left in Testing as a repeating invalid refresh token. | A saved credential can still need a reconnect click when the token stops refreshing. Community threads treat that as auth recovery, not a mapping rewrite. | Reconnect is only available for connections you own. The Zap owner must have access to every app connection used in the Zap. A Gmail connection can expire even after a Google sign-in to Zapier. |
| Who gets notified | Default emails: an error that blocks a successful run, and an automatic disable. You can also turn on warning emails. Make sends immediately, then a digest (5 minutes for errors, 15 minutes for warnings). | No default error inbox. Wire the Error Trigger to email or Slack. It runs on automatic failures, not on a manual test. You can’t test that path with a canvas Run click. | Default: email the account address immediately. You can switch to hourly summary or Never. Autoreplay waits until the final retry fails before sending mail. A custom error handler sends no error email. |
| False-green vs hard fail | Skip marks Success even when a bundle was dropped. Resume writes a substitute and still says Success. A scheduled scenario can look on while AccountValidationError has already stopped the run. Consecutive errors, quota, or AccountValidationError are hard stops. | A Success execution is not a checked destination write. A stored credential that no longer refreshes can sit in the list while live runs fail. An unset error workflow means nobody is paged. Self-host adds the box to own. | On stays on while a search Safely halted, a run sits On hold, or Autoreplay is Scheduled. Handled error keeps the Zap on. Unhandled Errored can turn the Zap off if 95% of runs error in the last 7 days. |
n8n Self-host uses the same editor. If you will own Docker, backups, and 2am, read When n8n self-host is worth it — losing the encryption key is not a reconnect.
OAuth reconnect ownership
A dead token is an auth miss on the failure taxonomy — reconnect, do not rewrite the flow. The product still names who can click that button.
- Make. OAuth grants access for a limited time. Reauthorize; if that fails, create a new connection. AccountValidationError happens during initialization and cannot be caught by an error handler. A Make Community thread names a Google Cloud OAuth app left in Testing as a repeating “invalid refresh token.” Sources: Make — connect an application · Make — fix errors and warnings · Make Community — connection expires every 7 days / invalid refresh token · accessed 2026-09-17
- n8n Cloud. The credential can remain saved after the access token stops refreshing. Community threads describe a reconnect click as the recovery — not a mapping rewrite. Self-host adds the encryption key — that is a different job on When n8n self-host is worth it. Sources: n8n Community — OAuth2 token won’t refresh · n8n Community — generic OAuth2 API not refreshing the token · accessed 2026-09-17
- Zapier. Reconnect is only available for connections you own. When you publish a Zap, the Zap owner must have access to every app connection used in it. A Zapier Community Gmail thread reports “This account connection is expired” after a Google sign-in — reauthorize Gmail, not only the Zapier login. Sources: Zapier — manage your app connections · Zapier — Zap owner must have every connection · Zapier Community — Gmail reconnect / account connection expired · accessed 2026-09-17
Name who clicks reconnect on Who should own your automations. If the prompt keeps coming back, use When automations look fine — do not invent an eighth class.
Who gets notified
A handler only helps if someone is actually alerted. Vendor cloud pages the vendor for the runtime. The team still pages a person for a failed write.
- Make emails errors and deactivations by default. Warning emails are optional — they cover issues that need attention, including connection problems. Make sends immediately, then a digest (5 minutes for errors, 15 minutes for warnings) with a link to scenario history. Source: Make — manage your email preferences · accessed 2026-09-17
- n8n does not mail unless you wire it. Create a workflow that starts with the Error Trigger, then select it under Error workflow in the job’s settings. You can reuse one error workflow across jobs. It does not run on a manual test. Sources: n8n — handle errors gracefully · n8n — Error Trigger · accessed 2026-09-17
- Zapier emails the account address immediately unless you change frequency to hourly summary or Never. Autoreplay withholds that mail until the fifth retry fails. A custom error handler sends no error email — the run is Handled. Never is not recommended on the vendor page. Sources: Zapier — manage notifications when errors occur · Zapier — set up custom error handling · Zapier — what is replay · accessed 2026-09-17
False-green vs hard fail
A green icon is not a destination write. A hard fail stops the job or pages someone. Confirm the last row, message, or file in the destination system.
- False-green. Make Skip or Resume Success after a dropped or substituted bundle. Zapier On with Safely halted, On hold, or Scheduled Autoreplay. n8n Success without a destination check, or a stored credential that no longer refreshes.
- Hard fail. Make Error (Rollback default), AccountValidationError, or a quota stop. Zapier unhandled Errored — a Zap turns off if 95% of its runs error in the last 7 days. n8n Failed with no Error Trigger wired.
Name the class on the failure taxonomy before you add steps. Catch the quiet miss on When automations look fine. If the stop is allowance, compare units on Same job, three meters — this page does not invent a monthly dollar.
Source: Zapier — how to troubleshoot errors in Zap workflows · accessed 2026-09-17
Then pick the tool
Then pick the tool
Design the failure path, then pick the product. The live card is Make vs n8n vs Zapier. Totals are not published yet. Honest cases: When Make is the right pick · When Zapier is still the right pick · When n8n self-host is worth it. Four questions: Which automation tool for small teams. When-to-pick: Make vs n8n vs Zapier — when to pick each. Scoring rules: Methodology.
Product links
Sources
Primary vendor error, retry, notification, and connection pages, plus community threads where this page names a reconnect loop. Access date for every citation on this page: 2026-09-17. We do not invent a reply count, a quote statistic, or a monthly dollar.
- Make — overview of error handling · accessed 2026-09-17
- Make — error handlers · accessed 2026-09-17
- Make — retry error handler · accessed 2026-09-17
- Make — incomplete executions · accessed 2026-09-17
- Make — automatic retry of incomplete executions · accessed 2026-09-17
- Make — fix errors and warnings · accessed 2026-09-17
- Make — manage your email preferences · accessed 2026-09-17
- Make — connect an application · accessed 2026-09-17
- Make Community — 429 rate limit / wait and try again · accessed 2026-09-17
- Make Community — connection expires every 7 days / invalid refresh token · accessed 2026-09-17
- n8n — handle errors gracefully · accessed 2026-09-17
- n8n — Error Trigger · accessed 2026-09-17
- n8n — Stop And Error · accessed 2026-09-17
- n8n — view all executions · accessed 2026-09-17
- n8n Community — OAuth2 token won’t refresh · accessed 2026-09-17
- n8n Community — generic OAuth2 API not refreshing the token · accessed 2026-09-17
- Zapier — review run statuses in Zap workflows · accessed 2026-09-17
- Zapier — manage notifications when errors occur · accessed 2026-09-17
- Zapier — set up custom error handling · accessed 2026-09-17
- Zapier — how to troubleshoot errors in Zap workflows · accessed 2026-09-17
- Zapier — what is replay · accessed 2026-09-17
- Zapier — replay Zap runs · accessed 2026-09-17
- Zapier — manage your app connections · accessed 2026-09-17
- Zapier — Zap owner must have every connection · accessed 2026-09-17
- Zapier Community — Gmail reconnect / account connection expired · accessed 2026-09-17
FAQ
What does “success” mean in Make, n8n, and Zapier?
Read the vendor status, not only the On switch. Make Skip and Resume can end as Success after a module error. n8n filters Success on the Executions list — that is a completed run, not a checked write. Zapier Successful is a completed path; Safely halted is a purposeful non-match; Handled error means an alternative workflow ran. Name the class on the failure taxonomy before you rewrite the flow.
Do these tools retry failed runs automatically?
Sometimes. Make retries ConnectionError, RateLimitError, and ModuleTimeoutError when incomplete executions are on — they are off by default. A Retry handler can auto-complete later. n8n retries from the Executions list; there is no account-wide Autoreplay. Zapier Autoreplay retries Errored steps up to five times and waits to email until the last attempt fails. It does not replay On hold or Safely halted. Custom Zapier error handlers turn Autoreplay off for that Zap.
Who owns the OAuth reconnect?
A named person. Make: reauthorize or replace the connection — AccountValidationError cannot be handled. Zapier: only the connection owner can click Reconnect, and the Zap owner must have every connection. n8n: a saved credential can still need a reconnect click. Name that owner on Who should own your automations. If reconnect loops, use When automations look fine — do not invent an eighth class.
Who gets notified when a job fails?
Make emails errors and deactivations by default; warnings are optional. Zapier emails the account address immediately unless you change frequency — Autoreplay and custom handlers suppress that mail until (or unless) a hard fail remains. n8n does not mail by default; you must wire an Error Trigger. If nobody is named, the job can fail quietly.
How is a false-green run different from a hard fail?
A false-green run still looks healthy — Make Skip/Resume Success, Zapier Safely halted or On, n8n Success without a destination check. A hard fail stops or pages: Make Error after Rollback, quota, or AccountValidationError; Zapier Errored (and a high error ratio can turn the Zap off); n8n Failed with no error workflow. Confirm the destination write. Classes stay on the failure taxonomy.
How is this page different from the failure taxonomy and When automations look fine?
This page compares what each product calls success, who retries, who reconnects, and who is emailed. The failure taxonomy is the class list. When automations look fine is the operator checklist for reconnect loops and silent breakage. This page does not reprint that table or that checklist.
Does this guide change the Make vs n8n vs Zapier scores?
No. Totals are not published yet. It is not a fourth cloud seat. Then pick from When Make is the right pick, When Zapier is still the right pick, or When n8n self-host is worth it.
Affiliate disclosure
Make Partner (pc=stackscorecard) is live elsewhere on the site and is not the primary call here. n8n and Zapier stay official-site URLs with UTM only — no invented IDs and no affiliate ID. Rankings are not sold. Policy: Affiliate disclosure · Disclosures.