Skip to content
Tutorials Center
My learning

Topics

Lesson 12 of 36 ยท Zaps, Triggers and Actions

Zap History, Errors, Retries and Replay

Read Zap History, understand run statuses and common errors, use error notifications, auto-replay and manual replay, and build safe retry logic with runnable Python stand-ins.

Automations fail. An app goes down for ten minutes, a password expires, a field arrives empty. What separates a good setup from a bad one is how quickly you notice, how easily you diagnose, and whether you can recover without creating duplicates. This lesson covers Zap History, error handling and replay.

Zapier cannot run on this site. Status names and features can change; we describe common behavior in words and recommend checking Zapier's help pages. The Python blocks are labelled stand-ins modeling retries and replays.

Zap History

Zap History (sometimes called Runs) lists every run with a status, the Zap name, time and often a link to details. Open a run to see, per step, the data received and the data sent or returned, plus any error message. Typical statuses (names may differ slightly):

Status Meaning
Success All steps ran.
Filtered / Stopped A filter or a "no results" condition ended the run on purpose.
Error / Halted A step failed and the run stopped.
Delayed / Waiting The run is paused (a delay step or waiting for an event).
Held / Throttled The run was held, for example because of rate limits or account issues.
Replayed A rerun of an earlier run.

Use the search and filters (by Zap, status, date) to find what you need. Data retention is limited, so note important findings.

Common error causes

  1. Expired or revoked connection: reconnect the account.
  2. Missing required field: the source sent an empty value.
  3. Validation errors from the target app: a bad email, an invalid dropdown value, text too long.
  4. Rate limits: too many requests in a short time.
  5. App outage or timeout: temporary; usually works on retry.
  6. Permission problems: the connected user cannot access that folder or record.
  7. Changed app: a field was renamed or deleted since you built the Zap.

Temporary problems (outages, timeouts, rate limits) are good candidates for retry. Permanent ones (bad data, missing permission) need a human fix.

Error notifications

By default, Zapier can email the Zap owner when a run errors, and can send digests. Check your notification settings: if alerts go to a person who left, nobody notices. For critical Zaps consider adding a team alert. A second, scheduled Zap can also check your own tables for items that never finished.

Auto-replay and manual replay

Auto-replay, where available, automatically retries runs that failed because of certain temporary errors, after a delay. Manual replay lets you rerun a failed run from the history after you fix the cause. Replaying uses the original trigger data, so it reproduces the same event. Two cautions:

  • If steps before the failure already did their work (created a record), replaying from the start may do it again. Design with find-or-create so repeats are harmless.
  • Replays consume tasks like any run.
flowchart TD
  R["Run fails"] --> K{"Temporary error?"}
  K -->|"yes"| A["Auto-replay after a delay"]
  K -->|"no"| H["Notify owner"]
  H --> F["Fix data, access or Zap"]
  F --> M["Replay from history"]
  A --> OK{"Success?"}
  OK -->|"no"| H
  OK -->|"yes"| D["Done"]

Temporary failures retry on their own; permanent ones need a person, then a replay.

Retry with backoff: a stand-in

Retrying immediately often fails again. Waiting longer each time (backoff) gives a struggling service room to recover. The stand-in simulates a service that fails twice, then works.

Python Stand-in: retries with exponential backoff

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Replay without duplicates

This is the most important idea of the lesson. A replay re-sends the original event. If the first run already created an invoice, a naive replay creates a second one. The stand-in shows a replay-safe step using the order ID as an idempotency key.

Python Stand-in: replay-safe invoice creation

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Debugging method

When a run fails:

  1. Open the run and find the first failing step.
  2. Read the error message in full; copy it.
  3. Compare the data received with what the step expected.
  4. Check the connection and the target app's own status.
  5. Fix the cause; then replay (or re-test in the editor).
  6. If it will happen again, add a guard (filter or path), a default, or an alert.

Monitoring over time

  • Skim history weekly for patterns: the same step failing each Monday?
  • Track the success rate and number of errors of critical Zaps.
  • Document known quirks in the Zap's description.
  • Keep contact details of app owners to ask about outages.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What does a manual replay use?

    Replay reruns the same event, so steps that already worked might repeat their effects.
  2. 2. Which error is best for automatic retry?

    Temporary problems often disappear; permanent problems need a human fix.
  3. 3. How can you make replays safe?

    Checking a stable identifier before creating prevents duplicates on repeat.
  4. 4. Why wait longer between retries (backoff)?

    Immediate retries tend to hit the same failure; growing delays help recovery.

Summary

  • Zap History shows runs, statuses and per-step data; it is the main debugging tool.
  • Temporary errors suit retry or auto-replay; permanent ones need a human fix.
  • Make sure error notifications reach someone who will act.
  • Replays reuse the original data, so design actions to be repeat-safe.
  • Debug from the first failing step, then add a guard so it does not recur.

Next: Connections and Authentication