Managing your instance
Open Instances in the left sidebar to see every instance your team has. Click any row to open its detail page - everything below lives there.
Instance status
The status badge tells you where your instance is in its lifecycle:
| Status | What it means |
|---|---|
| Provisioning | We are setting up the server and installing the software. Nothing for you to do. |
| Awaiting host | BYOS only - we are waiting for you to confirm the hostname before we can start. |
| Live | Running and reachable. |
| Stopped | The app process is stopped, but your data is still there. |
| Restarting | A restart was requested and is in progress. |
| Failed | Provisioning did not complete. Contact support with the instance ID. |
The page refreshes automatically - you do not need to reload it to watch progress.
Restart
A Restart button appears on live instances. Use it when n8n is unresponsive, after changing an environment variable, or when a workflow is stuck in a state it should not be in.
A restart does not affect your data, your workflows, or your credentials. Execution history is preserved. The instance is unreachable for roughly 30-60 seconds during the cycle.
Restart brings the app process down and up again on the same container. It does not re-provision the server, re-apply configuration, or update the app.
Changing your app version
New n8n releases are installed inside the maintenance window you set on this page - or during SpliceRun's default hours if you have not set one. We check that no workflow is running, take a backup, and put the old version back automatically if the new one does not come up healthy.
Crossing a major version is the exception: that always waits for you to agree.
When a newer release is available, the instance page shows it. Update now installs it immediately rather than waiting for your window.
The history panel records every version change on that instance: what it moved from and to, when, and whether it succeeded.
Your window, the default hours, security releases and going back are all covered in Updates and maintenance windows.
Going back to the previous version
If a version change causes a problem - a workflow stops working, a credential breaks, a node disappears - you can return the instance to the version it was running before:
- Open the instance detail page.
- Choose Go back to <version>, under the version panel.
- Confirm.
Your workflows, credentials and execution history are not touched. Only the application version changes. Updates then pause on that instance for 30 days, so it does not move forward again while you are still investigating - security releases still apply.
The option relies on the backup taken immediately before the change, so it is available for 7 days. It returns you to the version immediately before the most recent change, not several. If you have already gone back, or changed version again since the problem appeared, open a support ticket rather than repeating it.
SSL certificate
Your instance comes with a TLS certificate managed by SpliceRun. The instance page shows its status and expiry date. Certificates renew automatically - you do not need to do anything.
If the SSL status shows a problem, it is almost always a DNS issue. See Custom domain SSL troubleshooting.
Execution timeouts
n8n has a built-in execution time limit. If your workflows are hitting it regularly, the instance page surfaces a timeout alert:
- How often timeouts are happening
- Which workflows are most affected
- Whether an upgrade (more CPU or memory) is likely to help, or whether the workflow itself needs to be redesigned
Timeouts cut executions short - they do not crash the instance or affect other workflows running at the same time. Why is my n8n slow? has more on diagnosing the cause.
Request limits: upload size and how long a request may take
These are limits of the proxy in front of your instance, not of n8n, and they are the ones people hit first when a workflow handles files or calls a slow API. On instances we host:
| Limit | What hits it | |
|---|---|---|
| Request body size | 100 MB | A webhook posting a file, a form upload, an API call with a large payload |
| How long one request may take | 1 hour | A workflow that responds to a webhook only after doing a lot of work |
| Time to connect | 15 minutes | Rarely relevant |
Both of the first two changed on 14 August 2026. Before that the body limit
was the web server's stock 1 MB, which rejected larger uploads with a
413 Request Entity Too Large before they ever reached n8n, and a single
request was cut at 15 minutes.
Two things worth knowing:
- Going over the body limit looks like a client error, not ours. The sender
gets
413and n8n never sees the request, so nothing appears in your execution list. If a webhook seems to vanish only when the payload is large, this is the first thing to check. - The request limit is not the execution limit. A workflow can run far longer than an hour when it is triggered by a schedule rather than by an HTTP request — nothing is waiting on the other end of a connection. The hour applies to how long a caller will be held open. See Execution timeouts for the separate, n8n-side limit.
These limits come from the proxy configuration on the machine. On a bring-your-own-server instance that machine is yours, so its own limits apply.
Workflow failures
A red alert on the instance page means n8n is reporting persistent execution errors. The panel shows the failing workflows and the error message so you can find the problem without opening n8n.
This is a summary of errors n8n itself reported - it does not cover every possible failure mode, only ones that reach n8n's execution log.
Backups on the instance page
The instance detail page shows the last backup timestamp and the most recent restore points available. You can trigger a restore directly from here. Full details are in Backups and recovery.
Cancelling or discarding an instance
- Before provisioning is complete: use the Discard button to abandon the request and stop billing.
- After provisioning: use Cancel subscription in the billing settings. Your instance stays live until the end of the current period, then is deprovisioned. See Cancellation and data policy.