Help for owners
My n8n workflows stopped after an update because a community node broke
Short answer
Community nodes are installed on disk separately from n8n itself. On Docker, recreating the container or upgrading can remove them unless the nodes folder is kept, and the node's own developer can release a breaking update. If an n8n Cloud instance will not start, the owner can disable all community nodes in the Cloud console and restart.
Does this sound like your situation?
- n8n shows a missing packages error after a restart or an upgrade.
- A workflow that used a special node now shows an unknown or red node.
- The instance will not start since I installed a community node.
- The node is still listed but behaves differently after an update.
- Nobody remembers who installed the node or from where.
Why does this happen?
Community nodes live on the disk, not inside n8n
n8n's troubleshooting page says community nodes are installed onto the disk and have to be available at startup. If they are not, n8n shows a missing packages error.
Confirmed by docs.n8n.io
Recreating a Docker container can remove them
The same page says that recreating the container or upgrading n8n can remove the packages. Persisting the nodes folder is the recommended fix. An alternative setting reinstalls missing packages at startup, but it can slow startup and cause health checks to fail.
Confirmed by docs.n8n.io
The node's author can break it
n8n's risks page says node developers may release updates that break existing functionality, and that depending on how the developer versions the node, upgrading can break every workflow that uses it.
Confirmed by docs.n8n.io
Unverified nodes are unreviewed code
Installing from npm means running unverified code from a public source. n8n states such nodes have full access to the machine n8n runs on and to data in your workflows. Verified nodes are those n8n inspects against data and system security requirements.
Confirmed by docs.n8n.io
Cloud and self-hosted behave differently
Installing from npm is only available on self-hosted instances, and unverified community nodes are not available on n8n Cloud. A workflow built on a self-hosted instance with an unverified node cannot simply be moved to Cloud.
Confirmed by docs.n8n.io
What can I do about it today?
- Stop changing thingsDo not keep restarting, reinstalling or upgrading. Each change can hide what happened.
- Write down the node and the errorTake a screenshot of the error, the node's name and the n8n version. That is what anyone helping will ask for first.
- Find out how n8n is hostedCloud, Docker on a server or something else? The recovery differs. Ask whoever set it up.
- On n8n Cloud, use the documented safe modeIf the instance will not start, the Cloud console has an Instance settings option, Disable all community nodes, then Save and restart. It appears only when Verified community nodes is on. If the instance is offline, n8n says you can sign in with a magic link.
- Export your workflows before experimentsIn n8n, download the affected workflows as files so a copy exists outside the instance.
- List which workflows need the nodeA short list of what the business loses while it is broken helps decide what to fix first.
When should I bring someone in?
- The instance will not start and you cannot reach the settings.
- The node's developer has stopped maintaining it.
- The workflows move money, orders or customer data.
How do I handle it?
- Identify how the instance was builtVersion, hosting, volumes and which nodes are installed, from the server itself rather than from memory.
- Back up before touching itWorkflows, credentials and the node folder are copied first, with a restore I have tried.
- Restore the node in a persistent placeFor Docker I make sure the nodes folder survives container recreation, and pin the node's version.
- Decide: keep, replace or remove the nodeIf it is unverified or abandoned, I look for a built-in node or an HTTP request step that does the same job.
- Test on a copy before upgradingUpgrades go through a staging copy, with the workflows that matter run once before production.
The price depends on what I find, so I quote in writing after a short discovery call. Tell me what is happening and I reply within 24 hours.
What do owners ask about this?
Is it safe to install any community node?
n8n warns that unverified nodes run unreviewed code with full access to the machine and your workflow data. Install only ones you trust.
Can I turn community nodes off?
Yes. n8n documents a setting for self-hosted instances and an Instance settings option on Cloud.
Will upgrading n8n fix it?
Not necessarily. In Docker, an upgrade can be what removed the packages.
Where do I find the steps for my version?
In the n8n documentation for your own version. Ask before changing a production instance, and take a copy of the workflows first.
Sources
How was this page made?
I drafted this page with AI assistance from the official documents listed under Sources. Found a mistake? Tell me.