Picture a scheduled n8n workflow that worked during testing, then appears to stop after a change. The workflow is still visible, the nodes look correct, and no obvious error appears. In many cases, the problem is not the business logic. The workflow was never published, the trigger is using a different time zone than expected, or the execution is waiting on a deployment, credential, or event condition.
n8n workflows not running usually require two checks first: confirm the workflow is published and confirm that its trigger uses the intended time zone. A workflow that is inactive will not respond to production triggers, even if manual testing succeeds. A scheduled workflow can also run at a different local time because n8n evaluates the schedule using the workflow or instance time zone, depending on the deployment and configuration. After those checks, review the trigger event, credentials, execution history, environment variables, and downstream node conditions.
This guide explains how publishing and time zones affect n8n workflow execution, how to distinguish a trigger problem from a logic failure, and how to troubleshoot n8n Cloud and self-hosted environments without changing several variables at once.
Why n8n Workflows Stop Running After They Worked
An n8n workflow has several separate states that are easy to treat as one. The workflow can be saved, manually tested, active for production events, connected to valid credentials, and deployed in a running n8n environment. Those states are related, but they do not mean the same thing.
A manual test proves that n8n can execute the selected path under the conditions supplied at that moment. It does not prove that a production trigger is active, that a webhook is reachable, or that a schedule will fire at the expected local time.
The most common distinction is between manual execution and production execution:
Manual execution starts when a user selects a test option or runs a workflow from the editor.
Production execution starts when an active trigger receives an event, reaches its schedule, or responds to an external request.
A workflow that is not active or published may still be testable in the editor.
A workflow that is active can still fail because of credentials, network access, trigger configuration, or node logic.
This is why changing a downstream node first is usually the wrong move. If the trigger never creates an execution, the problem exists before the first business action runs.
A useful troubleshooting record separates what you observed from what you assume:
| Observation | What it proves | What it does not prove |
|---|---|---|
| Manual run completes | The selected path worked with the test input | The production trigger is active |
| Workflow appears in the editor | The workflow is saved and accessible | The deployed version is current |
| No execution appears | The trigger did not create a visible execution, or logging and retention need review | The workflow logic is broken |
| A red execution appears | Some part of the run failed | The trigger or schedule is configured correctly |
| Webhook responds in a browser | The endpoint may be reachable | The sending application delivered the expected request |
| Scheduled run occurs at the wrong hour | A time zone or schedule interpretation differs | Every node uses the same time zone |
That separation prevents a common failure pattern: editing credentials, code, and node expressions before confirming that the workflow is actually eligible to run.
If the issue involves a broader deployment, integration, or workflow design problem, our n8n workflow automation services cover the technical areas that sit around individual trigger troubleshooting.
Is the n8n Workflow Published or Active?
The first production check is simple: open the workflow and verify its active or published state. The exact label and interface behavior depend on the n8n version and hosting model, but the underlying question remains the same: is this workflow enabled to respond to real events?
Saving a workflow does not necessarily activate it. A draft can contain a valid trigger and pass a manual test while remaining disconnected from production event handling. This is particularly easy to miss after importing a workflow, duplicating an existing workflow, or making a small edit and leaving the workflow inactive.
Check these conditions before changing the workflow logic:
Confirm that the intended workflow, rather than a duplicate or older version, is open.
Verify that the workflow is active or published for production use.
Save recent edits before activating it.
Confirm that the trigger node is configured for production behavior.
Generate a new test event after activation instead of relying on an earlier manual run.
Publishing also matters after changes. A workflow may have an active production version while the editor contains newer unsaved or unpublished changes. The editor view can therefore look correct even though the running version does not contain those changes.
For webhook workflows, distinguish the test webhook URL from the production webhook URL. Test URLs are intended for listening while the workflow is being manually tested. Production URLs depend on the workflow being active and reachable through the deployed n8n instance. Sending a request to the test endpoint after closing the listening session does not demonstrate that the production workflow works.
For schedule triggers, activation is equally important. A schedule node does not run merely because the workflow contains a cron-like rule. The workflow must be enabled in the environment that is expected to execute it, and the n8n process must be running.
How Do n8n Time Zones Affect Scheduled Workflows?
n8n time zones affect when scheduled triggers fire and how date-based expressions are interpreted. A schedule such as “9:00 AM” is incomplete unless the workflow or instance time zone is known. A team in New York, a server configured for UTC, and a user viewing the workflow from California may all interpret the same business requirement differently.
For US teams, write the requirement in an unambiguous form during troubleshooting. For example, “run at 9:00 AM Eastern Time every weekday” is more useful than “run at 9 AM.” If the workflow serves multiple countries, record the IANA time zone identifier, such as America/New_York, rather than relying only on an abbreviation like EST. IANA identifiers account for daylight saving time transitions, while fixed abbreviations do not fully describe seasonal clock changes.
Time zone troubleshooting should examine more than the schedule node. Review:
The workflow time zone, if configured.
The n8n instance time zone and deployment configuration.
The operating system or container time zone for self-hosted installations.
The time zone used by the source application.
The time zone attached to incoming date values.
The time zone used by expressions, filters, and database queries.
Daylight saving time behavior for the relevant region.
A schedule can fire at the expected moment while a later filter excludes the data. For example, a workflow scheduled near midnight might retrieve records using a UTC date boundary, while the business expects a local calendar day. The trigger runs, but the search returns no records, creating the appearance that the workflow did not run.
Use a controlled test time when reviewing schedules. Avoid testing only around daylight saving time changes or at midnight, where date boundaries create additional ambiguity. Log the execution start time, the intended local time, the resolved UTC time, and the date range sent to the downstream application. That evidence shows whether the problem is the trigger schedule or the query window.
Versich also supports n8n automation developer services for teams that need help reviewing trigger behavior, custom API requests, deployment settings, credentials, and execution logging across an existing automation environment.
Check the Trigger Type Before Debugging the Nodes
The trigger type determines what “not running” means. A schedule trigger waits for the n8n process and scheduler. A webhook trigger waits for an HTTP request. An application trigger waits for an event from an integration or polling mechanism. A manual trigger waits for a person to start the workflow.
Treating all trigger failures as schedule problems leads to wasted testing. Identify what should have started the workflow, then test that event directly.
| Trigger type | What should start it | First diagnostic test | Common hidden issue |
|---|---|---|---|
| Schedule Trigger | A matching date and time | Wait for the next controlled interval or temporarily use a near-term schedule | Workflow or server time zone differs from the business time zone |
| Webhook | An HTTP request to the production endpoint | Send a controlled request with a unique event ID | Test URL used instead of production URL, or workflow is inactive |
| App event trigger | A new or changed record | Create one clearly identifiable test record | The trigger polls on an interval or filters out the event |
| Manual Trigger | A user action | Run the workflow from the editor | Manual success does not validate production behavior |
| Queue or event-driven deployment | A queued execution and available worker | Check queue, worker, and execution status | Main process is available but workers are stopped or misconfigured |
Webhook testing should include a unique identifier, a timestamp with an explicit offset, and only the minimum data needed for the test. That makes it easier to locate the request in execution history and avoids confusing a replay with a new event.
Application triggers deserve special attention because some integrations do not receive events instantly. Polling triggers check at intervals and may store a cursor, timestamp, or last-seen value. If the source system timestamps records in UTC while the workflow compares them with local time, valid records can fall outside the retrieval window.
A slightly overlapping retrieval window is safer than a narrow window with no duplicate protection. If a workflow polls every five minutes, a small overlap helps account for clock skew and delayed source updates. Idempotency keys, unique event IDs, or destination-side duplicate checks then prevent the overlap from creating duplicate records.
What Should You Check in n8n Execution History?
Execution history tells you whether the workflow failed before the first node, failed inside a node, completed with unexpected data, or never started at all. Start there before editing the workflow.
Search for executions around the expected run time and compare the result with the trigger type. A scheduled workflow should produce evidence near its scheduled window. A webhook workflow should produce an execution after the request reaches the production endpoint. A polling trigger may show a successful run that found zero qualifying records.
The execution status alone is not enough. Inspect the input and output around the first meaningful decision point. A green execution can still produce no business action when a filter, IF node, Switch node, or expression routes the item away from the intended branch.
A practical review follows this sequence:
Check whether any execution exists in the expected time window.
Confirm the execution belongs to the correct workflow and environment.
Identify the first node that received data.
Compare the received timestamp, event ID, and record count with the source system.
Inspect filters, conditions, and expressions that could intentionally stop the path.
Review the final destination state, not only the execution color.
Execution data can contain personal information, API responses, names, email addresses, and other sensitive fields. Limit execution access to the people who need it, define suitable retention settings, and avoid logging full payloads when a record ID is enough for diagnosis. Self-hosted deployments also require attention to database backups, encryption, environment variables, and administrative access.
If an execution is absent, focus on publishing, trigger delivery, scheduling, deployment health, and execution retention. If an execution exists but stops, focus on credentials, input data, node configuration, expressions, and downstream services.
Credentials, Environment Variables, and Deployment Health
A workflow can be active and correctly scheduled but still fail as soon as it reaches an authenticated node. Credential problems commonly appear after a credential is rotated, an environment is cloned, an OAuth connection expires, or a workflow is moved between n8n Cloud and self-hosted deployments.
Check the credential used by each external node, not just the first integration. A workflow that reads from one system and writes to another needs valid access on both sides. Custom HTTP Request nodes also require a review of the base URL, authentication method, headers, scopes, and request body.
Self-hosted n8n environments introduce additional dependencies:
The main n8n process must be running.
The database must be available and writable.
The scheduler or worker configuration must match the deployment mode.
Environment variables must be present in the running container or process.
Reverse proxy settings must allow webhook traffic.
TLS certificates and domain routing must be valid.
Queue mode requires healthy workers as well as the main process.
System clock configuration should be consistent across relevant services.
A restart can hide the underlying issue if it clears a temporary condition without correcting the configuration. Record the current status, relevant logs, and the time of the failed test before restarting. Afterward, run a controlled test and verify the resulting business state.
Deployment changes also explain why a workflow works in one environment and not another. Credentials, webhook URLs, environment variables, time zones, and external allowlists frequently differ between development, staging, and production. Compare those settings explicitly rather than assuming that an imported workflow is operationally identical.
A Practical Troubleshooting Matrix
The fastest diagnosis comes from matching the symptom to the layer most likely responsible. The following matrix helps narrow the search without changing unrelated parts of the workflow.
| Symptom | Most likely layer | What to verify next |
|---|---|---|
| Manual run works, production event does nothing | Publishing or trigger activation | Confirm active state, production URL, and event delivery |
| Scheduled workflow runs several hours late | Time zone configuration | Compare workflow, instance, container, and business time zones |
| No execution appears at all | Trigger, deployment, or retention | Check endpoint reachability, scheduler health, and execution history settings |
| Execution starts but receives zero items | Source query or polling cursor | Inspect date filters, last-seen values, and source-system timestamps |
| Workflow fails at an API node | Credentials or request configuration | Test token, scopes, URL, headers, rate limits, and response body |
| Workflow succeeds but no record is created | Branching or destination behavior | Inspect IF or Switch paths, destination response, and duplicate rules |
| Webhook works in test mode but not production | URL or active state | Use the production URL and confirm public routing and TLS |
| Workflow runs after restart only | Deployment or queue health | Review worker processes, database access, and process logs |
Do not make five changes between tests. Change one relevant condition, create a uniquely identifiable test event, and record the result. That method preserves causality and makes recurring issues easier to document.
How to Prevent Publishing and Time Zone Problems
Prevention starts with making operational assumptions explicit. A workflow should have a documented owner, trigger type, production endpoint or schedule, expected time zone, credential dependencies, and test method.
For scheduled workflows, document the local business time and the IANA time zone identifier. Include what should happen on weekends, holidays, daylight saving transitions, and missed runs. A schedule that sends a daily report needs a different recovery rule from a schedule that synchronizes records continuously.
For webhook workflows, document the production endpoint, authentication method, expected payload shape, retry behavior, and idempotency approach. The sending system should know whether it should retry a timeout, a 4xx response, or a 5xx response. n8n should be able to distinguish a repeated delivery from a new event when the source retries.
For polling workflows, define the retrieval window and duplicate-handling strategy. Store a stable source identifier where possible. Do not rely solely on a display name or a local timestamp to identify whether an item has already been processed.
A lightweight operational checklist includes:
Workflow name and owner.
Active or published status.
Trigger type and production endpoint.
Expected schedule and IANA time zone.
Source and destination credentials.
Retry and duplicate-handling behavior.
Execution retention and access rules.
Alert recipient and escalation path.
Recovery test and rollback procedure.
Error notifications should identify the workflow, execution time, failed node, record or event identifier, and a link or reference that helps an authorized operator investigate. A notification that only says “workflow failed” creates another manual investigation step and does not establish whether data was lost.
Our guide to n8n workflow error notifications covers the relationship between failure detection, ownership, investigation, and recovery. For business-critical automations, also test the destination state. A green execution does not prove that the expected record, message, approval, or downstream update exists.
When Should You Bring in n8n Troubleshooting Help?
Use internal troubleshooting when the issue is isolated, the deployment is understood, and the test data is safe to handle. Bring in specialist help when failures cross workflow logic, credentials, infrastructure, APIs, and operational controls.
A broader review is appropriate when:
Multiple workflows stopped around the same time.
A self-hosted deployment has inconsistent scheduler or worker behavior.
Production and test environments behave differently.
Webhook delivery depends on a reverse proxy, firewall, or private network.
Date and time logic crosses multiple regions.
The workflow processes sensitive information.
Duplicate prevention and recovery evidence are not documented.
The team cannot determine whether a missing result is a trigger failure or a downstream data issue.
A structured review should examine the workflow definition, active version, trigger behavior, execution history, credentials, custom code, API responses, deployment configuration, database health, and monitoring. The goal is not merely to make one execution turn green. The goal is to establish why the workflow stopped, whether any events were missed, and how the team will detect the next failure.
If you are ready to scope that kind of review, our n8n workflow rescue and troubleshooting support is relevant to workflows with failed executions, difficult deployments, custom integrations, or missing operational documentation. You can also contact our team with the workflow type, hosting model, trigger behavior, and the evidence already collected.
Conclusion: Take the Next Troubleshooting Step
When n8n workflows are not running, start with the boundary between the editor and production. Confirm the workflow is published, identify the real trigger, verify the production endpoint or schedule, and then compare every relevant time zone. Only after those checks should you move into credentials, node logic, API responses, and deployment health.
The most useful evidence is specific: an execution timestamp, a unique event ID, the resolved time zone, the first received item, and the final destination state. That evidence turns a vague “n8n stopped” report into a traceable technical problem.
Run one controlled test with the workflow active, the intended production trigger, and an explicitly documented time zone. If the result is still unclear, contact Versich with the execution history and deployment details so the next troubleshooting step starts with evidence rather than guesswork.
